Friends PC

[jerubedo]

23 hours ago, LogicWeasel said:

 Anyone who's seen AMD's track record with how often they change sockets can see they do it far less than Intel.  They said they're retaining AM4 until further notice and so far every report I've seen strongly believes their claim that they're supporting it into 2020.  You may suggest that you doubt this and you're welcome to think so, but I have far more reason to think that info is more credible than guessing at when Intel decides to arbitrarily cut off support for their last motherboard chipset to promote a new arbitrary chipset they "reinvented" .  (See: Z270 can't have CoffeeLake and Intel gives me the impression they're ready to say we'll support Z390 for newer CPUs "if we feel like it"). 

No, I was talking about it being pointless to debate on whether INTEL will change the socket with Icelake, not about AMD changing the socket. I have no doubt that they will keep the socket until 2020, but that doesn't mean it'll still be an upgrade path. You'll still need a new CHIPSET to take advantage of PCIe 5.0 coming later this year, and DDR5 (and USB 3.2). So while the socket may stay the same, your ability to use it for a new CPU that uses those new technologies will be hampered by needing a new chipset, meaning you'll need a new motherboard anyway. Which, by he way, is the same thing with Intel. Intel hasn't changed sockets in a while now. We've been on 1151 for ages now. The CHIPSET keeps changing.

 

23 hours ago, Stefan Payne said:

That's like 250W regions.

You need some real beefy heatsinks.

Not if you are overclocking and under-volting, which should be very possible for 4.8GHz. I just did that with my 9700K. I was able to get stable 4.8GHZ (a +200MHz bump on clock speeds from default) and with a voltage of 1.216 V. The default voltage my board was providing for the default turbo of 4.6 all-core was 1.232. So I got +200MHz with MINUS 0.016 V. And yes, it was 24 hour Prime95 stable, AND 24 hour LinX stable. And for a bonus 24 hours OCCT stable.

[LogicWeasel]

15 minutes ago, jerubedo said:

Which, by he way, is the same thing with Intel. Intel hasn't changed sockets in a while now. We've been on 1151 for ages now. The CHIPSET keeps changing.

I fail to see how it is the same thing with Intel.  You say it's been ages, but just 4 years ago Intel Broadwell used LGA 1150 , a whole 1 pin count difference, what a socket that was.  At least when AMD changes chipset without a socket change, a bios update is all that may be needed to use a new CPU in an older chipset motherboard with the same socket.  Intel sees a chipset change as an excuse to zone-out new CPUs from using older boards while also simultaneously blocking backward compatibility so older CPUs can't pair with the newer features (I was amused by the example set by the Chinese H310 motherboard Linus made a video about, which confirmed it was possible to run a 7th Gen in a chipset meant for 8th Gen, neat little find).

[jerubedo]

11 minutes ago, LogicWeasel said:

  You say it's been ages

It has. It's been 4 years. That's an eternity in the tech world. In June 2015, Skylake came out on 1151, then came Kaby Lake, then Coffee Lake, then Coffee Lake refresh.

 

11 minutes ago, LogicWeasel said:

At least when AMD changes chipset without a socket change, a bios update is all that may be needed to use a new CPU in an older chipset motherboard with the same socket.

That won't be the case with big physical hardware changes like PCIe 5.0 and DDR5. PCIe 5.0 will need a PCIe 5.0 slot on the board, and other physical changes to the board as well. The same goes with DDR5. DDR5 will still use the same 288 pins, but there will need to be physical changes to the board to as well. 

 

11 minutes ago, LogicWeasel said:

Intel sees a chipset change as an excuse to zone-out new CPUs from using older boards while also simultaneously blocking backward compatibility so older CPUs can't pair with the newer features

I'm not denying that. What I am saying is that as it stands today, both AMD and Intel will have relatively the same level of upgradeability because of PCIe 5.0, DDR5 RAM, and USB 3.2. In a couple of years we'll all need new motherboards to support those features regardless of which brand we choose today.

[Stefan Payne]

9 hours ago, jerubedo said:

Yes, of course in some limited games AMD will win.

No, its not that limited and its not AMD its 4C/4T vs. 6C/12T.

And its the same crap as 15 years ago, when people like you claimed that Single Cores are better for gaming - just to be proven wrong a year or two later.

And the same shit happened a year or two later with the 4 Core CPUs. People claimed that the 4 Core Chips are better for gaming, just to be proven wrong again.

 

How the heck can you claim that 4C/4T is enough today, when there already are people complaining about 100% Load on 4C/8T CPUs?!

 

That, my dear, is the future. In some shitty threadded games or games that have a low CPU load, that might be the case but those are getting less and less.

Especially when the next gen Consoles gets some pretty nice/beefy CPUs (7nm Zen2 with unknown clockrates but I'd expect something around 3,5GHz -compared to around 2GHz 8 Core Jaguar we have right now.

 

9 hours ago, jerubedo said:

You've found a few examples, but still in the majority of games, Intel wins.

right now, maybe.

BUT:
Where it counts the 6C/12T CPU AMD wins over the 4C/4T Intel CPU because of higher CPU Load.

Again what you are talking about is not CPU Performance, its CPU Latency.

So when the Load on the CPU increases it looks like AC: Origins and other games.

 

9 hours ago, jerubedo said:

And no, not all of them are high FPS situations. Look at Arma III. On the Ryzen system there are dips below 60, and not on the i3.

ARMA 3 is a single Thread Game, that doesn't even use more than 2 Thrads IIRC.

It doesn't matter if you have a 2Core CPU or a 16 Core for that Game.

 

You really need to look into WHY that happens and n ot just take it as is without any other infos.

 

And its a really really old game as well.

Look at more modern games from last and this year, not from some shit someone did 10 years ago and released in 2013.

 

Assasin's Creed Origins and Odyssey are the future, they load all 8 Cores and 16 Threads very heavily. That will be more common once the new console generation comes out.

You can than kiss the "Intel is better!!!11" trope goodby...

But you can already do that right now.

There are already people complaining that their 4C/8T System runs like shit in some Multi Player Shooter. 

We just had a guy saying his CPU was constantly at 100% and that his Game doesn't run too well. IIRC it was Apex Legends.

 

9 hours ago, jerubedo said:

The same goes for Assassin's Creed Origins. The i3 gets 53 FPS average and the 2600 gets 46. That's a huge lead for i3. There are more examples as well if you care to look them up.

Where did you get that stuff from?!
And in the Link I've provided your beloved i3 is at 48/63fps while the 1600x is at 60/74fps.

Yours runs barely, the 1600x is a bit whole lot better.

That's +30% more.

 

And what do you say to the people Who are already complaining about their Intel 4C/4T CPU running pretty badly, while it runs well on his friends Ryzen 1600/1700 or 2600/2700???

 

"you're wrong, you must be imagining things"?? 

Or what?

9 hours ago, jerubedo said:

This is pointless. It's all rumors. Neither you nor I know. We'll just have to wait and see.

The Leadership of the Company knows.

The Leadership of the Company (AMD) said that it will not change sockets unti DDR5.

The Leadership of the Company (Intel) never said anything about the sockets that I know of.

Some Rando on the Internet is not relevant. What the Leadership says on Stage/Events is relevant.

 

And the facts are what we said. You are free to not believe it but when the Leadership of a company says something about a product, it can have some legal rammifications as you could sue them if it does not happen for malicious reasons.

9 hours ago, jerubedo said:

Yes people like you or me might upgrade, but not your average buyer who is content with performance as is.

That is your take, not the reality.

The average user looks at upgrading when the system runs like shit and causes problems.

YOUR system instantaneously causes Problems.

It will get worse with time.

The AMD System is the better one without a doubt.

And you can just drop in a better CPU with just a BIOS Update.
THe CPU Prices will drop.

The CPU Prices on your plattform will not drop. A 7700K for the last 1151 Socket is still more than a similar CPU on a different Plattform cost with a Board.

Its rediculously expensive.

 

9 hours ago, jerubedo said:

You aren't keeping up with the news, then. PCIe 5.0 is coming THIS YEAR. See here: https://www.tomshardware.com/news/pcie-4.0-5.0-pci-sig-specification,38460.html

And you should look at real sites, not some that have some obvious bias. Just look at the bashing they got for the "JUST BUY IT!" stuff...

And look a the dates of that stuff as well!!

The Date for PCIe 1.0 was around 2002 or so, first product came with i915 shitset back in around 2004.

Don't have a date for PCIe 2.0 as I didn't care about that.

But PCIe 3.0 has a date of 2010. First product was in 2012.

PCIe 4.0 is dated at 2017 or so. First products was VEGA 20 but no CPU to really support it.

 

So its the Specification, not products. And that can take another couple of years, if it ever happens...

And that can be limited to Server/Workstations as well.

As there is already talk about some real problems/rammifications with PCIe 4.0 and the max. possible lenth of traces.

Some knowledgable people are talking about PCIe 4.0 only possible for the first PCIe and not for multiple PCIe ports on the Motherboard.

 

So PCIe 5.0 might not really be something we will see in the Desktop market soon, if at all.

And PCIe 4.0 isn't even here yet...

And PCIe 5.0 might be limited to Server Stuff with high frequency Mezzanine Connectors.

 

Something like that:

https://www.samtec.com/connectors/high-speed-board-to-board

 

PCIe 4.0 is already at around 8GHz...

 

Oh and to make your view worse:
BIOS Support for unnamed "Upcoming AMD Processor" dropped a couple of days ago...

For example my MSI B450I already has a BIOS with that stated:

https://www.msi.com/Motherboard/support/B450I-GAMING-PLUS-AC

- Support new upcoming AMD cpu.
 

[jerubedo]

32 minutes ago, Stefan Payne said:

No, its not that limited and its not AMD its 4C/4T vs. 6C/12T.

And its the same crap as 15 years ago, when people like you claimed that Single Cores are better for gaming - just to be proven wrong a year or two later.

And the same shit happened a year or two later with the 4 Core CPUs. People claimed that the 4 Core Chips are better for gaming, just to be proven wrong again.

 

How the heck can you claim that 4C/4T is enough today, when there already are people complaining about 100% Load on 4C/8T CPUs?!

 

That, my dear, is the future. In some shitty threadded games or games that have a low CPU load, that might be the case but those are getting less and less.

Especially when the next gen Consoles gets some pretty nice/beefy CPUs (7nm Zen2 with unknown clockrates but I'd expect something around 3,5GHz -compared to around 2GHz 8 Core Jaguar we have right now.

All you keep doing is talking about the future, but this has literally been the argument for YEARS now. Yes, SOME games are finally taking advantage of more cores and threads, but it is still only a few and it is not the overall trend. Most games still prefer frequency and IPC gains rather than cores and threads. People have been saying, "Games will use more cores and threads 'soon'" for roughly 7 years or so now. When things actually change and using cores and threads is the general trend, then my stance will change. And when that change finally does come, I can almost guarantee it won't be the case of games using as many cores/threads as is available. It'll likely be 8 threads used, and then we'll be stuck there for another 4-8 years. But again, speculation.

 

32 minutes ago, Stefan Payne said:

You really need to look into WHY that happens and n ot just take it as is without any other infos.

It doesn't matter WHY it happens. All that maters is that it does happen. Right now. In the present.

 

32 minutes ago, Stefan Payne said:

Where did you get that stuff from?!
And in the Link I've provided your beloved i3 is at 48/63fps while the 1600x is at 60/74fps.

The link you gave is for ODYSSEY, not ORIGINS. I'm talking about Origins. i3 wins. See here (look at 0:55):

 

 

32 minutes ago, Stefan Payne said:

The Leadership of the Company knows.

The Leadership of the Company (AMD) said that it will not change sockets unti DDR5.

The Leadership of the Company (Intel) never said anything about the sockets that I know of.

Some Rando on the Internet is not relevant. What the Leadership says on Stage/Events is relevant.

Please keep up, I wasn't talking about AMD, I was talking about Intel. You said that it's unlikely that Icelake will work on Z370/Z390, then I said it's pointless to argue because it's all rumors. The fact that Intel hasn't said anything literally means nothing. If they actually said, "no, Icelake will not be supported on Z390" that would be a different story.

 

32 minutes ago, Stefan Payne said:

THe CPU Prices will drop.

You have no way to know that.

 

32 minutes ago, Stefan Payne said:

And you should look at real sites, not some that have some obvious bias. Just look at the bashing they got for the "JUST BUY IT!" stuff...

And look a the dates of that stuff as well!!

Toms' is still a reputable source of news in the tech world.

 

32 minutes ago, Stefan Payne said:

So its the Specification, not products. And that can take another couple of years, if it ever happens...

And that can be limited to Server/Workstations as well.

 

Did you actually read the article? It says right there: "The PCI-SIG expects to ratify the final 1.0 revision in the first quarter of 2019, and the first PCIe 5.0 devices should debut this year. Broader availability should come in 2020."

 

32 minutes ago, Stefan Payne said:

Oh and to make your view worse:
BIOS Support for unnamed "Upcoming AMD Processor" dropped a couple of days ago...

For example my MSI B450I already has a BIOS with that stated:

https://www.msi.com/Motherboard/support/B450I-GAMING-PLUS-AC

- Support new upcoming AMD cpu.

I never said they won't support Ryzen 3000. That was literally never an argument. I said the likelihood of actually being able to support 4000 in the wake of PCIe 5.0 and DDR5 isn't possible unless they won't adapt those technologies in favor of keeping their socket/chipset. But if that's the case, that'll give Intel a huge lead, so I doubt they'd do that. 

[Stefan Payne]

55 minutes ago, jerubedo said:

All you keep doing is talking about the future, but this has literally been the argument for YEARS now.

No, I'm talking about NEW and upcoming Games, as well as experience from the past.

YOU'RE WRONG. Deal with it!

More Cores and Threads are better.

55 minutes ago, jerubedo said:

Yes, SOME games are finally taking advantage of more cores and threads, but it is still only a few and it is not the overall trend.

Yeah, same arguments I've heard time and time again, when the Dual and Quad Cores came out.

Now Intel vs. AMD. 

It wasn't true then, why the hell should it be true now that the lower core count should be the better part?!
 

That makes no sense...

And we have people complaining that games run like shit on their 4 Cores. You ignore that. WHY?

55 minutes ago, jerubedo said:

Most games still prefer frequency and IPC gains rather than cores and threads.

And how many of those games run well on a 10 Year old DELL?!
How many of those really need the performance?

In the end, you're claiming stuff that is just not true. The "IPC" is a term you don't really seem to understand. IPC is not that different. The real Difference is the Latency. ANd if you do not need latency but performance, AMD wins.

How many cases do you need??

Does someone have to remember you and flame and bash you for the stuff you're saying a couple years later?

 

55 minutes ago, jerubedo said:

People have been saying, "Games will use more cores and threads 'soon'" for roughly 7 years or so now.

Yes and now its true.

There are examples of that.

Assasin's Creed Odyssey and a couple of other games.

 

The mainstream more than 4 Core CPUs are only a bit over 2 Years old, before that, most people had 2 Core CPU and some 4 Cores - as they were rediculously expensive. Remember: I7-920 was under 200€ at one time.

 

55 minutes ago, jerubedo said:

It doesn't matter WHY it happens.

Yes, it does. You NEED to know why something happens to know what to do against that and how to deal with it.

 

55 minutes ago, jerubedo said:

The link you gave is for ODYSSEY, not ORIGINS. I'm talking about Origins. i3 wins. See here (look at 0:55):

Ryzen 2600 vs i3 8100 Tested in 15 Games + Rendering Benchmarks

5,078 views

How long did you look for that Video?! 

Sorry, but you're cherry picking some shit that proves your point.

 

How about this:
i3-8100 is very similar to i7-7600k. Both are same CPU Architecture, both have 4 Cores and 4 Thread. THe 7600K is even clocked a bit higher (=has a boost).

So now look here:

https://www.techspot.com/article/1525-assassins-creed-origins-cpu-test/

https://www.computerbase.de/2017-10/assassins-creed-origins-benchmark/3/#diagramm-cpu-tests-in-assassins-creed-origins-auf-einer-gtx-1080-ti-strix-1920-1080

 

In the last one, the i5-7600K is about the same as a Ryzen 3/1300X. 

Here, the Ryzen 1600x is, again, 30% faster than the i5.

 

So you're cherry picking a scene that shows what you want to believe.

 

But here another one:

https://gamegpu.com/mmorpg-/-онлайн-игры/anthem-test-gpu-cpu
Anthem, Ryzen 1600X slightly faster than i3-8100

 

Devil May Cry 5, although very playable with both, the 1600x annihilates the 8100:

https://gamegpu.com/action-/-fps-/-tps/devil-may-cry-5-test-gpu-cpu

We're talking about 100/136 vs. 140/181

40% higher min FPS, 33fps higher average.

 

Similar behaviour in Resident Evil 2 Remake:

https://gamegpu.com/action-/-fps-/-tps/resident-evil-2-test-gpu-cpu

109/144 for your i3, 140/173fps for the 1600x.

 

The Division 2, your i3-8100 gets beaten by an FX8350!

https://gamegpu.com/mmorpg-/-онлайн-игры/the-division-2-beta-test-gpu-cpu

63/79fps for the 8100.

86/113 fps for the 1600x.

+36% for low FPS, 43% on average.

 

Anthem both are the same.

 

Atlas:

https://gamegpu.com/mmorpg-/-онлайн-игры/atlas-test-gpu-cpu

1600x slightly faster with 70/80fps vs. 62/72fps for the i3/8100

The 7600k is a good amount faster. So that is that.

 

Shadow of the TOmb Raider, DX12, 8100 is about the same as the Ryzen 2400G, the 1600x is faster (80/101 for the 1600x vs. 56/86fps for the i3)

https://gamegpu.com/action-/-fps-/-tps/shadow-of-the-tomb-raider-test-gpu-cpu-2018

 

The well "beloved" Battlefield V also has the 8100 behind and only be slightly faster than the 4 COre Ryzen 2400G.

Here we're talking about 78/101fps vs 97/127fps for the Ryzen 5/1600x.

https://gamegpu.com/action-/-fps-/-tps/battlefield-v-test-gpu-cpu-2018

 

In far Cry 5, both are about the same.

 

Kingdom Come Deliverance, 1600x: 45/63 vs 37/52 for your i3/8100

https://gamegpu.com/rpg/ролевые/kingdom-come-deliverance-test-gpu-cpu-2018

 

Just Cause 4, the 1600x also is slightly faster and gets to the limit, i3 has 4fps lower min FPS:

https://gamegpu.com/action-/-fps-/-tps/just-cause-4-test-gpu-cpu

 

Dark Siders 3 both about the same.

 

Sunset Overdrive, you guessed it, Ryzen 5 beats the i3:
113/133fps for the 1600, 89/107 for the i3/8100
https://gamegpu.com/action-/-fps-/-tps/sunset-overdrive-test-gpu-cpu

 

Hitman 2, you guessed it, Ryzen 5 faster again (even if it is just slightly):
https://gamegpu.com/action-/-fps-/-tps/hitman-2-test-gpu-cpu

 

So the claim that Intel is better always and in every price is just not true.

 

55 minutes ago, jerubedo said:

Toms' is still a reputable source of news in the tech world.

No, they are a joke in the circles that know a bit more.

Or rather they are seen as Shills for the big companies. 
The "Just Buy it" (google it!) Proves it.

They were shunned by everyone in the Tech world.

 

Here a video about Toms Hardware's "Just Buy it" stuff from a more reputable source:

 

They were a joke 15 years ago, when they did the AMD vs. Intel Livestream stuff - wich spectacularly backfired as the Intel Board died, while AMD did it without any incident. Even then the "people in the know" made fun of them.

 

55 minutes ago, jerubedo said:

Did you actually read the article? It says right there: "The PCI-SIG expects to ratify the final 1.0 revision in the first quarter of 2019, and the first PCIe 5.0 devices should debut this year. Broader availability should come in 2020."

Yes and PCIe 4.0 Devices were demonstrated like 1,5 years ago or so.

But the first product with PCIe 4.0 for Consumer was released just last year...

It took more than a year for it to be implemented.

It took around 2 Years for PCIe 3.0 to be implemented.

So even if PCie 5.0 was released right now, there is a delay in it to be implemented in actual products.

AMD is more flexible than the others as they have a dedicated I/O Die but it makes no sense to touch it just for PCIe without anything else...

Its very unlikely that we will see that combined with DDR4 SDRAM.

 

55 minutes ago, jerubedo said:

I never said they won't support Ryzen 3000. That was literally never an argument. I said the likelihood of actually being able to support 4000 in the wake of PCIe 5.0 and DDR5 isn't possible unless they won't adapt those technologies in favor of keeping their socket/chipset. But if that's the case, that'll give Intel a huge lead, so I doubt they'd do that. 

Pls click on the Links I've provided and read that stuff what I've posted.

I've shown you the design of the upcoming Ryzen 3000 Series Chip.

 

To be more clear:
They put the Northbridge beside the CPU. The CPU is connected with an interface to the I/O Chip.

 

The Socket and Memory is dependant on the I/O Die.

The I/O Die connects to the CPU die.

 

They literally use the same CPU in Ryzen 3000 and Rome and probably Consoles. Its the same CHip.

The I/O die is different each time (and the amount of CPU Dies).

 

AMD Can support Zen x on AM4, if they want to as the CPU Die is only the CPU and the Socket Compatibility is due to the I/O Die, not the CPU!

That's comparable to the early Intel Mobile Sockets that had the Chipset on that module.

Or how the ASUS P/I-P65up5 works.

The CPU Card thing is integrated onto the Am4 Package with Matisse.

They have a Chipset Die and up to 2 other dies (its speculated that there will be 1 CPU and 1 GPU Die as well as 2 GPU Dies).

Here a Picture of the Board:

https://electrelic.com/electrelic/node/153

And also the earlier Pentium I and Pentium PRO CPU Board. There was also a (rare!) Pentium 2 CPU Board (ASUS C-PKND its called).

[jerubedo]

9 hours ago, Stefan Payne said:

No, I'm talking about NEW and upcoming Games, as well as experience from the past.

YOU'RE WRONG. Deal with it!

More Cores and Threads are better.

Nope. Look at these popular 2018 releases. NEW GAMES. I can provide a video for any ones you ask for:

 

i3 8100 wins:

Far Cry 5
Just Cause 4
Shaodw of the Tomb Raider
PUBG
Black Ops 4
Dragonball Fighter Z

 

Functional tie:

Monster Hunter World
Hitman 2

Anthem

 

Ryzen 5 2600 wins:

Assassin's Creed Odyssey
Forza Horizon 4
Battlefield V

 

9 hours ago, Stefan Payne said:

Yeah, same arguments I've heard time and time again, when the Dual and Quad Cores came out.

Now Intel vs. AMD. 

It wasn't true then, why the hell should it be true now that the lower core count should be the better part?!

It's actually still kind of true now. A dual core Pentium G4560 beats the Ryzen 3 1200. See here and skip towards the end where he swaps the Athlon with the Ryzen 3:

 

 

9 hours ago, Stefan Payne said:

How about this:
i3-8100 is very similar to i7-7600k. Both are same CPU Architecture, both have 4 Cores and 4 Thread. THe 7600K is even clocked a bit higher (=has a boost).

That's not apples to apples, and therefore irrelevant. The 1600x is a higher clocked CPU than the 2600 and is higher priced.

 

9 hours ago, Stefan Payne said:

So you're cherry picking a scene that shows what you want to believe.

No, the 2 videos I've linked show both the in-game benchmark and an actual gameplay scene. Nothing cherry picked. Just some guy playing the game on both. @Stefan Payne, you're the one cherry picking by using the 1600X, which is not comparable to the 2600. The 1600X boosts to 4.0GHz and is more expensive than the 2600 which only boosts to 3.9GHz. The 1600X is $181 and the 2600 is $165, which I hope I don't need to point out that that's cheaper than the $118 i3. If you want to compare something to the 1600X, you'll need to compare to the either the i3 8350K or the i5 8400.

 

9 hours ago, Stefan Payne said:

Shadow of the TOmb Raider, DX12, 8100 is about the same as the Ryzen 2400G, the 1600x is faster (80/101 for the 1600x vs. 56/86fps for the i3)

Maybe in DX12 mode, but DX11 runs at a higher framerate overall in SOTTR, and i3 wins. Also stop comparing to the 1600x. We're talking about the i3 8100 vs the R5 2600. See here:

 

 

9 hours ago, Stefan Payne said:

In far Cry 5, both are about the same.

Nope:

 

9 hours ago, Stefan Payne said:

Just Cause 4, the 1600x also is slightly faster and gets to the limit, i3 has 4fps lower min FPS:

Maybe the 1600X beats it, but not the 2600:

 

9 hours ago, Stefan Payne said:

Hitman 2, you guessed it, Ryzen 5 faster again (even if it is just slightly):

No, they are a functional tie as noted above.

 

9 hours ago, Stefan Payne said:

Yes, it does. You NEED to know why something happens to know what to do against that and how to deal with it.

Not for end-consumers. It doesn't matter to them. All that matters is which performs better in the games they want to play, not why. Irrelevant argument.

 

9 hours ago, Stefan Payne said:

Devil May Cry 5, although very playable with both, the 1600x annihilates the 8100:

https://gamegpu.com/action-/-fps-/-tps/devil-may-cry-5-test-gpu-cpu

We're talking about 100/136 vs. 140/181

40% higher min FPS, 33fps higher average.

 

Similar behaviour in Resident Evil 2 Remake:

https://gamegpu.com/action-/-fps-/-tps/resident-evil-2-test-gpu-cpu

109/144 for your i3, 140/173fps for the 1600x.

Of course these 2 perform better on AMD hardware. They are AMD optimized games, which is why they're being given away with AMD hardware. The same goes for the Division 2.

 

9 hours ago, Stefan Payne said:

Ryzen 2600 vs i3 8100 Tested in 15 Games + Rendering Benchmarks

5,078 views

How long did you look for that Video?! 

Sorry, but you're cherry picking some shit that proves your point.

It took like 10 seconds. It doesn't matter how long it took, either. Data is data. And again, you keep comparing to 1600X, a more expensive CPU with a higher Boost clock than the 2600. You're the one not comparing apples to apples and cherry picking.

 

9 hours ago, Stefan Payne said:

The Socket and Memory is dependant on the I/O Die.

The I/O Die connects to the CPU die.

Right and that I/O die is not equipped to handle DDR5 speeds, so a motherboard change will still be needed to take advantage of DDR5 once it hits. Specifically the Data Flow Management System integrated circuit will need to be updated.

 

9 hours ago, Stefan Payne said:

The "IPC" is a term you don't really seem to understand. IPC is not that different. The real Difference is the Latency.

 

Lower latency MEANS higher IPC, whether it be lower latency from CPU cache or lower latency from the RAM. If data isn't being held back by higher latency then more instructions are able to be retired per cycle! That's the very definition of IPC. Therefore lower latency = higher IPC. Actually Anadtech said that same thing: "Improvements to the Cache Hierarchy: Lower Latency = Higher IPC" 

 

https://www.anandtech.com/show/12625/amd-second-generation-ryzen-7-2700x-2700-ryzen-5-2600x-2600/3

 

By the way, it's not only me who's saying that Intel is better for gaming right now. Within the past year, every major tech Youtuber has stated such including Jay, Linus, Steve, and Paul.

 

9 hours ago, Stefan Payne said:

Yes and PCIe 4.0 Devices were demonstrated like 1,5 years ago or so.

But the first product with PCIe 4.0 for Consumer was released just last year...

It took more than a year for it to be implemented.

It took around 2 Years for PCIe 3.0 to be implemented. 

So even if PCie 5.0 was released right now, there is a delay in it to be implemented in actual products.

AMD is more flexible than the others as they have a dedicated I/O Die but it makes no sense to touch it just for PCIe without anything else...

Its very unlikely that we will see that combined with DDR4 SDRAM.

 

https://www.extremetech.com/computing/250640-pci-sig-announces-plans-launch-pcie-5-0-2019-4x-bandwidth-pcie-3-0

 

"A quick progression from PCIe 4.0 to 5.0 might result in consumer hardware not shipping PCIe 4.0 for very long"

 

https://www.techspot.com/news/78355-pcie-50-ready-before-pcie-40-can-launch.html

 

"According to the group that finalizes PCIe specifications, PCI SIG, a variety PCIe 5.0 devices are in deep development and are expected to debut in the server market late this year"

 

https://www.digitaltrends.com/computing/pcie-50-spefications-ratified/

 

"In the Apple community, AppleInsider is hopeful that Apple will adopt PCIe 5.0 standard on its upcoming redesigned Mac Pro that will debut some time this year."

 

9 hours ago, Stefan Payne said:

Does someone have to remember you and flame and bash you for the stuff you're saying a couple years later?

 

I've explicitly stated that I'm only going off of right now. No one knows what the future will bring. Maybe every release in 2020 will be super optimized and use all cores and threads. Maybe not. I don't have a crystal ball. I'm not going to gamble and say, "The Ryzen is better because it'll be better for future games." I'm going off of how things are now. I've explicitly said that once the overall trend changes (as in once a majority of games coming out are using all cores/threads) then I will change my stance. So no, you can't come back and flame me because I'm making no such claim that the i3 will still be superior later on. Maybe it will be, maybe it won't. None of us can guess what developers will do. There are certain things that we do know for the future and certain things we do not. An example of what we know is that more RAM and VRAM will be better. An example we don't know is if Ryzen 3000 will actually take the gaming crown, or if games will be more optimized for more cores/threads.

 

I've shown so many examples of the cheaper i3 8100 beating the R5 2600 in today's games in average FPS, and 1% lows. Yes, AMD has a few wins as well, but the majority still goes to Intel. And if you're going to link anything else, stop comparing to the 1600X. As I've stated before it's more expensive and has a higher boost clock out of the box. Show me i3 8100 vs R5 2600 only since that's what we are discussing. Even at that the R5 2600 is still more expensive than the i3. The 1600x beats the 2600 in Assassin's Creed Origins for example. See here. The 2600 gets an average FPS of 57 and the 1600x gets 62. That's a 9% gain on the 1600x :

 

 

And again, this is all for GAMING ONLY. Once we take productivity into account, Ryzen wins hands down. So all of the above is for those buying a PC for gaming and nothing else.

 

On 3/9/2019 at 12:52 AM, Stefan Payne said:

Look at how relatively big the step from Ryzen 1700x -> 2700x was

You mean the 3% increase in IPC? Or the 10% overall increase from mostly upping the clock speed? 

 

https://www.anandtech.com/show/12625/amd-second-generation-ryzen-7-2700x-2700-ryzen-5-2600x-2600/4

 

Oh, wait, that sounds familiar doesn't it? Sounds just like Intel. So what relatively big step are YOU talking about?

 

By the way, take a look at the conclusion of that article as well: 

 

"Any users that would like high single threaded performance, or high performing 1080p gaming using a mid-range GPU, then Intel’s Core i5-8400 is going to fit the bill.

For hardcore enthusiasts, running high-end graphics at 4K or like getting their general compute on, the Ryzen 2000-series is looking the best choice."

 

Which is EXACTLY what I always say. 

[Stefan Payne]

9 hours ago, jerubedo said:

"In the Apple community, AppleInsider is hopeful that Apple will adopt PCIe 5.0 standard on its upcoming redesigned Mac Pro that will debut some time this year."

What are you talking about?!
We don't even have PCIe 4.0 yet and don't know much how it affects the mainboard PCB and you come in here speculating about PCIe 5.0 that has NOTHING to do with this topic?! 

That quote above shows it best. They hope that Apple will adopt it...

When its said that with PCIe 4.0 its not even possible to get the signal from the CPU to the mid PCIe Slot and that the signallevels are a huge problem with conventional PCB Design, you want to increase the frequency further...

Just read up on it and come with facts, not speculation and whisful thinking.

 

With this high frequencys, even an insect sitting on the traces right next to it or a bit more dust might influence the signal so that it could cause Problems. Its really not as easy as you think it is.

 

And then there's also the Problem of the Driver Power.

That means that PCIe 4.0 will increase power consumption as well...

That will be even worse with PCIe 5.0...

 

 

And even if, it might also be that we won't see PCIE 5.0 in a desktop enviroment ever because its just not practical/vaiable or just too expensive to do so due to the signal integrity, max. trace lenth and other stuff as well...

 

PCIe 4.0 was specified back in 2017, right now, almost 2 years later, we see first products. With PCIe 5.0, wich isn't even final, it won't be much different...

So its far far away, with luck its around 2022 that the first implementations hit the market.

 

But even today, 7 years after PCIe 3.0 was introduced, we still have PCIe 2.0 lanes in Chipsets to components. That is still in use and PCIe 3.0 has NOT replaced 2.0 fully...

And for Consumer stuff, it doesn't matter too much anyway. For most people that is. the difference between 8 and 16 lanes are not that big in a single GPU Setup.

And that's THE ONLY THING that could benefit from the higher bandwith. 

Where it really matters is Server and Workstation stuff. Mainly HPC. For those that's enormusly important.

 

I really don't get why you're hyping this stuff when its so far away and not even certain that it will arrive at consumer products as no products with PCIe 5.0 are announced and won't be for a long time.

It might also be that when PCIe 5.0 hits the market, nobody cares because other stuff like Gen-Z and CCIX have replaced it in the important Markets...

 

9 hours ago, jerubedo said:

I've shown so many examples of the cheaper i3 8100

And now look at this thread. How far it has gone from the original topic and that there are many examples that show you are wrong.

 

Just because YOU don't want to admit that the Ryzen 1600/2600 is the better choice. You still claim you are right when you're not and the signs are on the Horizon, people already complain about 100% CPU Load with their 4 Core CPUs, people complaining about Stutter in Battlefield Games with 4Core CPUs.

We just had such a case a couple of days ago!

 

There are also a ton of games out there that benefit from more than 4 Cores, hell even 8 Threads...

And its getting more and more...

 

 

Another thing to keep in mind:
more than 4 Cores in Desktop are a new thing. That only happened 2 Years ago. That we already have some games that benefit from 6 and 8 Cores with 12 and 16 Threads is amazing and shows that there is a need for such things, that "the Industry" was waiting for it.

And that the days of glorified 4 Core "high end" CPUs are numbered, the future belongs to 8 and some day 16 Core CPUs. It might take another year or two but it will happen. That is for certain.

Even if you claim it is not the case. 
And the one who bought the AMD System is happy that he bought the AMD, can upgrade to Ryzen 3000 and probably 4000. And with luck they might even do more than that, if they want to.

 

While the Intel i3/8100 Buyer will curse the day he bought it because its a dead end, no upgrade that makes sense...

All the games he wants to play run like shit.

 

YOU tell the same stuff that people with Single Cores when Dual COres came out said. The same shit that people with Dual Cores said when Quad Cores came out. Yeah, don't buy quad cores, dual cores are so much better for gaming. ANd You won't quad cores anyway.

And also: Buy Intel Dual Cores, don't look at the AMD QUad Cores...

 

They were wrong. Over and over. And now you really claim the same...

Just look at the past to know what the future will bring!!

 

9 hours ago, jerubedo said:

Or the 10% overall increase from mostly upping the clock speed? 

When did Intel have 10% overall increase from one Generation to the next? on average.

Especially when you look at how well Sandy-Bridge held up and people are still using i7-2600K and other 2k and 3k Chips today...

10% isn't too bad for some minor changes, don't you think?

Especially since its only a slight improvement of the 14nm process.

 

Why not admit that its not a bad change? Why do you have to bash AMD for this???

Doesn't make sense...

 

And Ryzen 3000 is different because its a real Shrink!

And you can still use the same socket. The last time you were able to experience a shrink in a socket was LGA1156 -> Sandy vs. Ivy Bridge. And before that it was Bloomfield -> Westmere. But that one was only for the highest end, not for the average consumer (i7-970, 980 and Extreme 980X and 990X)...

[jerubedo]

2 hours ago, Stefan Payne said:

What are you talking about?!
We don't even have PCIe 4.0 yet and don't know much how it affects the mainboard PCB and you come in here speculating about PCIe 5.0 that has NOTHING to do with this topic?! 

That quote above shows it best. They hope that Apple will adopt it...

When its said that with PCIe 4.0 its not even possible to get the signal from the CPU to the mid PCIe Slot and that the signallevels are a huge problem with conventional PCB Design, you want to increase the frequency further...

Just read up on it and come with facts, not speculation and whisful thinking.

I have now linked FOUR reputable sources that say that PCIe 5.0 devices are in deep development. Believe what you want but everyone in the industry says they are coming and that 4.0 will be leapfroged. It's relavent because you brought up that AMD was superior for upgradeability but it's not. Both brands will need a board change in the coming year or 2 for DDR5 and PCIe 5.0.

 

2 hours ago, Stefan Payne said:

And for Consumer stuff, it doesn't matter too much anyway. For most people that is. the difference between 8 and 16 lanes are not that big in a single GPU Setup.

And that's THE ONLY THING that could benefit from the higher bandwith. 

5.0 brings much more needed extra lanes as well for more devices like NVME drives, or wifi/Bluetooth adapters. 

 

2 hours ago, Stefan Payne said:

There are also a ton of games out there that benefit from more than 4 Cores, hell even 8 Threads...

And its getting more and more...

No there aren't and that's my point. You pointed out a handful and it does not represent the landscape right now. The overwhelming majority still favor Intel. For every game that favors Ryzen there are 3 more that favor Intel. Once that changes so will my tune.

2 hours ago, Stefan Payne said:

YOU tell the same stuff that people with Single Cores when Dual COres came out said. The same shit that people with Dual Cores said when Quad Cores came out. Yeah, don't buy quad cores, dual cores are so much better for gaming.

Yes and back then, at the time they were saying that, single core with high IPC and frequency still did better in some games and it was able to still perform well for another year or 2 before wide spread adaptation of dual core coding. So at the time they were correct. That single core CPU still lasted long enough, and then it was time to upgrade the CPU anyway. It's the same story now. Right now i3 is still better. And it likely will remain better for quite a few upcoming titles. Maybe for a year or 2, we'll see. At that point maybe everything will be multithreaded and then I'll be saying go Ryzen instead. But right now it is what it is. 

 

2 hours ago, Stefan Payne said:

When did Intel have 10% overall increase from one Generation to the next? on average.

Especially when you look at how well Sandy-Bridge held up and people are still using i7-2600K and other 2k and 3k Chips today...

10% isn't too bad for some minor changes, don't you think?

Every single new release Intel is in the 7-8% range for increase. They are both bad. 8% or 10% is nothing to write home about. 

 

2 hours ago, Stefan Payne said:

Why not admit that its not a bad change? Why do you have to bash AMD for this???

Doesn't make sense...

I'm not. I'm saying what every single reputable source has said for the past year including Linus, Jay, Steve, Paul, Tom's, and Anandtech: Intel for gaming. Ryzen for everything else. I have no brand loyalty. If Ryzen takes the crown I will recommend it across the board. Period. 

 

 

[jerubedo]

2 hours ago, Stefan Payne said:

people already complain about 100% CPU Load with their 4 Core CPUs, people complaining about Stutter in Battlefield Games with 4Core CPUs.

You do realize I'm not actually advocating 4 cores across the board right? This whole thing started because you came in here and said that AMD was the better choice because of upgradeability and performance. Then I said that was wrong and that Intel wins at every price point for gaming. The i3 beating the 2600 was just an example. The i3 would only be used in a budget build where it would outperform the 2600 in the majority of games and it's cheaper. For the budget the OP had I recommended the i5 8400, a 6 core CPU. Given an even higher budget I'd recommend the I7 9700K, an 8 core CPU. 

[GoldenLag]

14 minutes ago, jerubedo said:

I'm not. I'm saying what every single reputable source has said for the past year including Linus, Jay, Steve, Paul, Tom's, and Anandtech: Intel for gaming. Ryzen for everything else. I have no brand loyalty. If Ryzen takes the crown I will recommend it across the board. Period. 

in todays market it makes literally no sense to buy the i3 8100. i hope you realize this. like i really hope so. 4 cores are on the way down, while the 4 core 4 threads 3rd and 4th gen i5 are starting to stutter and loose ground.

 

i am a highy value oriented person and if the 8100 made sense id recommend it, but it doesnt. for the same reasons pple dont recommend the 8350k etc. the 8400 was great 8 months ago with its exellent pricing, but today its expencive and a dead end product for upgrades. 

 

yes intel takes the crown for gaming, but unless you have more than 1400$ to spend for your system it doesnt make sense. same applies for CPU, mobo and Ram upgrades. ill give em one; if you need a CSGO frame machine, grab a 1660ti and a 9700k. 

 

21 minutes ago, jerubedo said:

I have now linked FOUR reputable sources that say that PCIe 5.0 devices are in deep development. Believe what you want but everyone in the industry says they are coming and that 4.0 will be leapfroged.

ok. please show me the device that will feed said lanes? or at least a paper product. PCIe 4.0 is allready being prepared for deployment in the forms of CPUs that can feed said lanes with no bottleneck. 

23 minutes ago, jerubedo said:

It's relavent because you brought up that AMD was superior for upgradeability but it's not

um, it is. being supirior doesnt mean its 100% perfect and with no flaws. its not perfect. but its supirior to any x86 product avavible to consumers

24 minutes ago, jerubedo said:

Both brands will need a board change in the coming year or 2 for DDR5 and PCIe 5.0.

board change is sqedualled for 2021 for AM4. AM4 technically doesnt need a board change. and while DDR5 is on the horizon, the first year will be slow deployment in the enterprise section of the market, and there is also no need to to get faster memmory, DDR4 is fast enough for 16 modern cores running at full speed. 

 

with no PCIe 5 products on the horizon, we dont actually know when that will be available to consumers nor server enterprise. and while PCIe 4.0 is nice to see, its no huge revelation for us Consumers. we can still run our top of the line GPUs on 8 PCIe 3.0 lanes. 

 

8 minutes ago, jerubedo said:

You do realize I'm not actually advocating 4 cores across the board right? This whole thing started because you came in here and said that AMD was the better choice because of upgradeability and performance. Then I said that was wrong and that Intel wins at every price point for gaming. The i3 beating the 2600 was just an example. The i3 would only be used in a budget build where it would outperform the 2600 in the majority of games and it's cheaper. For the budget the OP had I recommended the i5 8400, a 6 core CPU. Given an even higher budget I'd recommend the I7 9700K, an 8 core CPU. 

yeah..... no

 

for the given budget a 2200G or 2600 with a gtx 1660ti or 2060 is the path to go. the core i5 8400 was good 8 months ago, today its not. the i3 has no place in such a system and at a lower budget its too expencive. 

 

32 minutes ago, jerubedo said:

5.0 brings much more needed extra lanes as well for more devices like NVME drives, or wifi/Bluetooth adapters. 

you know how said lanes work?

 

well step 1. device has x lanes to enter. meaning regardless of PCIe revision it is going to take x lanes if given them. it also will run at the spec it was designed with. meaning a PCIe 1.0 device will run at PCIe 1.0 regardless of the lane entering it. 

 

to makes them work you need multiplex chips to split the lanes. and these things arent cheap. rarely present on lower end boards and usually only contains one in the chipset. which is allready included. 

 

while theoretically they can feed more devices. they need infrastructure around them to feed the extra devices. 

35 minutes ago, jerubedo said:

Every single new release Intel is in the 7-8% range for increase.

you mean re-release. we havent actually seen an actual increase in performance per core over 3 gens. 7 through 9. with each being based on the same process with the same coredesign. ok they made the die longer, thats about it. i mean the CPUs are top notch, but they are to the core the same CPUs. 

 

Zen 1 to Zen + was a node shrink (no die shrink because it is still really 14nm) with changes to the memmory controller. we got a little bit out of it. the only thing that actually changed was that stuff got cheaper. 

 

 

you know, buying 4 core CPUs for anything but the most budget gaming system is silly. and dualcores honestly in the realm where you should consider a new build. a older 2nd gen with a used GPU and a PSU is at that point miles better. 

[jerubedo]

6 hours ago, GoldenLag said:

in todays market it makes literally no sense to buy the i3 8100. i hope you realize this. like i really hope so. 4 cores are on the way down, while the 4 core 4 threads 3rd and 4th gen i5 are starting to stutter and loose ground. 

 

i am a highy value oriented person and if the 8100 made sense id recommend it, but it doesnt. for the same reasons pple dont recommend the 8350k etc. the 8400 was great 8 months ago with its exellent pricing, but today its expencive and a dead end product for upgrades. 

There's a place for everything. There's a place for the 8100, a place for the R5 2200g, a place for the Pentium G5420, etc. If I were talking about the R5 2200g then I'm sure you wouldn't have even commented. The i5s and i3s are only loosing ground in outlying games, not the majority. As long as games still are coming out that favor frequency over cores and threads then they are still very much relevant CPUs. As even Stefan pointed out, both the i3 8100 and the R5 2600 are over 60 FPS in almost all games. So how exactly is it an irrelevant product? Is it because it doesn't hit 60 and stutters in like 5 games? Both you and him are only thinking about the future as if you have a crystal ball. You seem to think that a switch has been flipped on the major developers now that we have higher core count parts, and that they'll now magically develop differently. Well, maybe, but maybe not. We have no way to know! Developers are still developing for the underpowered consoles we have today and thinking about PC secondarily for the most part. When the next gen consoles hit, THEN I'm sure developers will start to change, but even then it'll take a while for them to catch up.

 

6 hours ago, GoldenLag said:

um, it is. being supirior doesnt mean its 100% perfect and with no flaws. its not perfect. but its supirior to any x86 product avavible to consumers

No it's not, because by the time a consumer is ready to upgrade, CPUs will be out that use DDR5 and PCIe 5.0, which will require a board change on both systems. 

 

6 hours ago, GoldenLag said:

board change is sqedualled for 2021 for AM4. AM4 technically doesnt need a board change.

Yes it will need a board change. If Ryzen 4000 only supports DDR5 (which is quite a possibility since it's way more expensive to support two generations of RAM) then the chipset on the motherboard needs to support that, making a board change necessary. 2021 is right around when most end consumers would be looking to upgrade. No one is upgrading from Ryzen 1000 to 2000 or from 2000 to 3000. People in general will be happy with 2000 performance even after 3000 hits. They will most likely be looking to upgrade by 4000, at which point all of what I'm saying becomes relevant. By then, all CPU releases will assuredly be DDR5 only.

 

7 hours ago, GoldenLag said:

for the given budget a 2200G or 2600 with a gtx 1660ti or 2060 is the path to go.

Umm, did you just recommend a 4c/4t part after going on and on about how you should never do that? The 2200g is 4 cores and 4 threads. And the i3 beats it in every single game, even the ones where the 2600 took the lead.

 

7 hours ago, GoldenLag said:

the core i5 8400 was good 8 months ago, today its not.

In what world? 6 cores is just fine, and even for more multithreaded games coming out it will hold up.

 

7 hours ago, GoldenLag said:

you know how said lanes work?

Of course I do. The PCIe 5.0 specification specifies that there will be no less than 24 lanes, meaning we'll never see as low as 16 lanes again. And yes, cards right now only need an 8x 3.0 lane to be fully utilized but soon enough they will be using 16x. Being stuck on 16 lanes has sucked.

 

7 hours ago, GoldenLag said:

you mean re-release. we havent actually seen an actual increase in performance per core over 3 gens. 7 through 9. with each being based on the same process with the same coredesign. ok they made the die longer, thats about it. i mean the CPUs are top notch, but they are to the core the same CPUs. 

 

Zen 1 to Zen + was a node shrink (no die shrink because it is still really 14nm) with changes to the memmory controller. we got a little bit out of it. the only thing that actually changed was that stuff got cheaper.  

 

Call it whatever you want. It doesn't matter that the cores haven't changed. They've still been able to get 7-8% gains each time, while Ryzen was only able to eek out 10% on a die shrink??? How is that "good?" And I'm not bashing AMD here. I'm bashing both companies. 30% is a good generation over generation performance gain, which is what both companies should target, not a measly 10%.

 

7 hours ago, GoldenLag said:

you know, buying 4 core CPUs for anything but the most budget gaming system is silly. and dualcores honestly in the realm where you should consider a new build. a older 2nd gen with a used GPU and a PSU is at that point miles better. 

This is just one example, but JayzTwoCents recently recommended a budget build using the Pentium G4560, a dual core part (albeit 4 threads). It performed admirably and even beat out the 2200g in terms of stutter.

[GoldenLag]

10 minutes ago, jerubedo said:

If I were talking about the R5 2200g then I'm sure you wouldn't have even commented.

Depends entirely on budget. The budget space for the 2200G is tiny.  And the 8100 and 2200G doesnt occupy the same space. 

12 minutes ago, jerubedo said:

As long as games still are coming out that favor frequency over cores and threads then they are still very much relevant CPUs. As even Stefan pointed out, both the i3 8100 and the R5 2600 are over 60 FPS in almost all games. So how exactly is it an irrelevant product?

Pricing and competing products. That is why. In other words the market.

14 minutes ago, jerubedo said:

Developers are still developing for the underpowered consoles we have today and thinking about PC secondarily for the most part

And said consoles are becomming easy to port over by them using x86 CPUs made by AMD and practically identical GPUs using the GCN Macroarchitecture.

15 minutes ago, jerubedo said:

Both you and him are only thinking about the future as if you have a crystal ball. You seem to think that a switch has been flipped on the major developers now that we have higher core count parts, and that they'll now magically develop differently

We are allready seing a shift. If this was 2 years ago id agree with you, but there is allready a shift. Indietitles will continue to use singlecore or low core becuase its easier to develop. But its been that way since the first dualcore.

17 minutes ago, jerubedo said:

No it's not, because by the time a consumer is ready to upgrade, CPUs will be out that use DDR5 and PCIe 5.0, which will require a board change on both systems

Depends entirely on the consumer. DDR5 will at the earliest get adopted in 2020. With 2021 which is what im looking at. 

 

PCIe standards will come if needed. Currently 4.0 is the new thing with 5.0 still being some time out. 

21 minutes ago, jerubedo said:

it will need a board change. If Ryzen 4000 only supports DDR5 (which is quite a possibility since it's way more expensive to support two generations of RAM) then the chipset on the motherboard needs to support that, making a board change necessary

This is all assuming that in 2020 AMD will adopt DDR5. Which they can very much decide only to be a server thing. Or only workstation. They have memmory controllers on a seperate die. 

22 minutes ago, jerubedo said:

Umm, did you just recommend a 4c/4t part after going on and on about how you should never do that? The 2200g is 4 cores and 4 threads. And the i3 beats it in every single game, even the ones where the 2600 took the lead

Yep. The pricing area for the 2200G is small, but its there. Its only because of pricing. Which sadly due to pricing the 8100 doesnt fit in. Especially with the lackluster upgrade path. 

24 minutes ago, jerubedo said:

And yes, cards right now only need an 8x 3.0 lane to be fully utilized but soon enough they will be using 16x. Being stuck on 16 lanes has sucked

We will have to wait for doublenthe required throughput than the Titan V for that. Which will come at some point i suppose.

25 minutes ago, jerubedo said:

In what world? 6 cores is just fine, and even for more multithreaded games coming out it will hold up.

Again, pricing here is key. And hyperthreading is showing especially helpfull when extending the life of CPUs.

27 minutes ago, jerubedo said:

Of course I do. The PCIe 5.0 specification specifies that there will be no less than 24 lanes, meaning we'll never see as low as 16 lanes again

I mean. Another 4 lanes for the consumer plattform is allways appriciated. Because we have 20 lanes. Also why would the number of lanes be a part of the throughput specification?

29 minutes ago, jerubedo said:

They've still been able to get 7-8% gains each time, while Ryzen was only able to eek out 10% on a die shrink???

I mean k did specify there was no dieshrink. Its still the 14nm design and a 14nm based process.

Ill let it slide though as dieshrink is a common term.

31 minutes ago, jerubedo said:

And I'm not bashing AMD here. I'm bashing both companies. 30% is a good generation over generation performance gain, which is what both companies should target, not a measly 10%

Well with the lack of nodeshrinks, we should dial back expectations a bit. 

33 minutes ago, jerubedo said:

They've still been able to get 7-8% gains each time, while Ryzen was only able to eek out 10% on a die shrink???

Last time i checked they gained nothing except increasing the boost clocks and pushing what can be called 95 watt TDP. 

 

They gained 10% on a small nodeshrink based on the same node and a copy of the same design yes. Kinda "meh" but its what the node shrink could deliver. 

35 minutes ago, jerubedo said:

This is just one example, but JayzTwoCents recently recommended a budget build using the Pentium G4560, a dual core part (albeit 4 threads). It performed admirably and even beat out the 2200g in terms of stutter

G4560 is old. And while pretty OK at its MSRP. It sadly cant be found there. Meanwhile the 200GE sits around MSRP and is overclockable using an MSI motherboard. (Because AMD cucked up a chipset update). And even without the BIOS its a better pick because of pricing. 

 

Again pricing os very key here. The pentiums would be recommended over the 200GE if they could ever be found at MSRP.

 

Im not saying the CPUs are bad, but in face of competition they dont stack up in terms of pricing and performance. 

 

39 minutes ago, jerubedo said:

2021 is right around when most end consumers would be looking to upgrade

 

Good thing that is when AM4 ends and AM5 (presumably) starts. And intel rolling out 10nm desktop late 2020. Which also will adopt the new standards.

 

Its not as if these companies wont adopt what is reasonably avavible. And even if 4.0 is leapfrogged. Hardware that uses 4.0 can at least utilize that instead of 3.0 when 5.0 devices are plugged in. 

[Stefan Payne]

7 hours ago, jerubedo said:

I have now linked FOUR reputable sources that say that PCIe 5.0 devices are in deep development. Believe what you want but everyone in the industry says they are coming and that 4.0 will be leapfroged. It's relavent because you brought up that AMD was superior for upgradeability but it's not. Both brands will need a board change in the coming year or 2 for DDR5 and PCIe 5.0.

I'm not saying that PCIe 5.0 is not important overall.

I'm saying that it doesn't matter for Desktop Use.

And if you would actually read your sources, that's what some of them actually say.

For example, I read this one:

https://www.design-reuse.com/articles/45712/pcie-5-0-vs-emerging-protocol-standards.html

 

NOWHERE in that article are they ever talking about the Desktop Market. And that's what I was telling you as well.

Everything there is "Enterprise" aka Server and also the so called "HPC" Market. 

And there is a need for high bandwith. For example because the Internet bandwith is exploding.

But name one thing where we need it in the PC Desktop area we need it.

I can tell you one thing: AMD HBCC...

 

7 hours ago, jerubedo said:

5.0 brings much more needed extra lanes as well for more devices like NVME drives, or wifi/Bluetooth adapters. 

No, it doesn't!
It increases frequency/bandwith per lane

It increases power consumption because it needs more power to drive the frequency.

 

NVMe Drives are at 2GB/Sec for the most part, max. is 4GiB/Sec. And most Drives are far away from that.

 

PCIe 5.0 for WiFi/Bluetooth?! Are you kidding?! That's rather low bandwith stuff. Bluetooth is even fine with USB 2.0 it doesn't require much bandwith at all as its a near field thing for devices like gamecontrollers, sound devices... And WiFi is just at around (theoretically) 1300MBit/Sec. So around the same as a GBit LAN thing. PCIe 1.0 is fine for that.

Please, please read up on that stuff, what it means and tells you.

 

7 hours ago, jerubedo said:

No there aren't and that's my point. You pointed out a handful and it does not represent the landscape right now. The overwhelming majority still favor Intel. For every game that favors Ryzen there are 3 more that favor Intel. Once that changes so will my tune.

You're wrong.

Because you miss the how and why. YOU are favoring Intel, not the games you claim.

As @GoldenLag and I tried to tell you, there are many instances in wich the 4 Core CPUs are crapping out and cause stutter, especially in Multi Player games.

 

7 hours ago, jerubedo said:

Yes and back then, at the time they were saying that, single core with high IPC and frequency still did better in some games

Yes, in some, ancient games that only use one or two th reads.

Not in all games.

Especially not in modern games.

And in upcoming games.

 

8 Cores at Desktop is 2 years old.

And we already have games that bitchslap 4 Core CPUs with an 8 Core one.

And that after ONLY 2 Years with available Hardware!

 

Dude you realy have no idea what that means. And that you are totally wrong with your low core count CPU...

The only thing holding it back are the consoles right now. But this generation is on its way out, the next generation is already in development and possibly announced later this year or early next year!!

At that point, with a strong 8 Core probably 16 Thread CPU, there is no reason to hold back no more!!

 

ANd yes, Consoles are the primary development plattform right now...

7 hours ago, jerubedo said:

and it was able to still perform well for another year or 2 before wide spread adaptation of dual core coding.

...and that "another year or two" is now.

Because there is a demand for 6-8 Cores in Desktop for at least 5 years!

Haswell refresh should have been 6 Cores with SMT back in the day, Skylake at the latest.

And the software is made more and more for Multicore.

 

If your problem doesn't scale that well, you come up with other stuff you can do, that uses the resources - what Ubi started with Origins and continued on building for Odyssey. You run a simulation of the enviroment in the background for example.

You increase the physics calculation/details.

You make more stuff destroyable.

 

Consoles have 8 Cores since 2013 or so. That is what is developed on.

Next gen Consoles are said to ship next year.

GPU Power increases slightly, Bandwith increases a bit as well, memory increases but the real jump this time is in CPU Power!!

The CPU in all the Consoles right now is ~8 Years old, clocked at around 2GHz - for the Premium COnsoles.

The PS4 had only a 1,6GHz CPU. A 3GHz 8 Core Zen2 with SMT enabled is a huge leap and around 2-4x the Performance.

7 hours ago, jerubedo said:

Every single new release Intel is in the 7-8% range for increase. They are both bad. 8% or 10% is nothing to write home about. 

No, its not.
Its lower than that.

ANd mostly only because of Clockrates.

the "IPC" you love so much, hardly increases.

THe last big step was Haswell.

 

7 hours ago, jerubedo said:

I'm not. I'm saying what every single reputable source has said for the past year including Linus, Jay, Steve, Paul, Tom's, and Anandtech: Intel for gaming. Ryzen for everything else. I have no brand loyalty. If Ryzen takes the crown I will recommend it across the board. Period. 

You are wrong.

As they only look at the higher end and rarely to never compare lower end.

And there the Intel don't look too good and are just too expensive.

[GoldenLag]

4 minutes ago, Stefan Payne said:

ANd yes, Consoles are the primary development plattform right now...

@jerubedo

Forgot to mention Consoles are used to using many cores to spread the load because they are "underpowered".

 

With them using copies of Desktop CPUs. Adopting the sale usage to desktop is easier than ever. 

[Stefan Payne]

8 hours ago, jerubedo said:

You do realize I'm not actually advocating 4 cores across the board right?

You are advocating for 4 Cores where it hurts people the most.

People who don't have much money and can't afford higher end stuff and are stuck with at most 150€ for the CPU.

 

If you recommend a  CPU/Plattform they can live with for 2 Years, you're giving bad advise.

If you recommend a CPU/Plattform they can live with for at least 4 Years, you don't give bad advise.

Simple as that.

 

8 hours ago, jerubedo said:

This whole thing started because you came in here and said that AMD was the better choice because of upgradeability and performance.

Yes, wich is a true statement. And you will see that!

While Intel is out to make as much money they can, they can afford to change the Plattforms.

AMD just wants to sell stuff and not spend the R&D for a new plattform.

 

So obviously AMD Plattforms last longer because they support the same Plattform and not arbitarily, for no technical reason, throw the plattform away.

 

Like this:

 

8 hours ago, jerubedo said:

Then I said that was wrong and that Intel wins at every price point for gaming.

Because you are wrong here and should admit that in the long run, the AMD is better.

Where it counts, the AMD beats the shit out of your beloved i3.

Where it doesn't matter, your i3 is faster.

8 hours ago, jerubedo said:

For the budget the OP had I recommended the i5 8400, a 6 core CPU. Given an even higher budget I'd recommend the I7 9700K, an 8 core CPU. 

So only Intel.
Why not AMD??

What's wrong with AMD?

You want to spend 500€ for an i5??

[GoldenLag]

6 minutes ago, Stefan Payne said:

So only Intel.
Why not AMD??

What's wrong with AMD?


You want to spend 500€ for an i5??

This is where the 1200$ or more for intel atm "rule" applies. 

 

Because CPUs have their place in the budget brackets.  But some like the 8100 and 8400 are left out because another CPU have taken the bracket where they would have been.

[LogicWeasel]

ITT:  I find out that Intel favoritism can lead to walls of text that get countered by people who know what they're talking about @Stefan Payne and @GoldenLag who try in vain to return the conversation to logical PC build choices at a given budget rather than assumptions about how an overpriced Z370/Z390 platform ages better.

[jerubedo]

16 minutes ago, Stefan Payne said:

No, it doesn't!
It increases frequency/bandwith per lane

It increases power consumption because it needs more power to drive the frequency.

 

NVMe Drives are at 2GB/Sec for the most part, max. is 4GiB/Sec. And most Drives are far away from that.

 

PCIe 5.0 for WiFi/Bluetooth?! Are you kidding?! That's rather low bandwith stuff. Bluetooth is even fine with USB 2.0 it doesn't require much bandwith at all as its a near field thing for devices like gamecontrollers, sound devices... And WiFi is just at around (theoretically) 1300MBit/Sec. So around the same as a GBit LAN thing. PCIe 1.0 is fine for that.

Please, please read up on that stuff, what it means and tells you.

You need to work on reading comprehension. I wasn't saying that Wifi and Bluetooth need extra bandwidth. I was talking about the extra 8 lanes that the PCIe-5.0 specification brings. Instead of 16 lanes minimum 5.0 defines a minimum amount of 24 lanes. Plugging in an M.2 drive and a wifi adapters brings your available lanes down to 8 for your GPU, which is enough for right now, but not the future. With 24 lanes we can have an M.2 drive, a wifi adapter, AND a GPU running at 16x. NVME drives will also get faster to take advantage of the extra bandwidth.

 

19 minutes ago, Stefan Payne said:

NOWHERE in that article are they ever talking about the Desktop Market. And that's what I was telling you as well.

Everything there is "Enterprise" aka Server and also the so called "HPC" Market. 

And there is a need for high bandwith. For example because the Internet bandwith is exploding.

Of course they are coming to server markets first. That's almost always the case. Sure that one article didn't mention desktop parts, but read the other articles, too:

 

"Desktop PC users will probably have to wait until 2020 to get their hands on PCIe 5.0 NVMe SSDs and GPUs"

 

23 minutes ago, Stefan Payne said:

You're wrong.

Because you miss the how and why. YOU are favoring Intel, not the games you claim.

As @GoldenLag and I tried to tell you, there are many instances in wich the 4 Core CPUs are crapping out and cause stutter, especially in Multi Player games.

No there aren't There are a handful of games that are that such case. That's a drop in the bucket compared to the thousands of games available on the market.

 

25 minutes ago, Stefan Payne said:

Yes, in some, ancient games that only use one or two th reads.

Not in all games.

Especially not in modern games.

And in upcoming games.

Of course not in modern games, but for games of the time they were still correct. A single core still beat dual core in many games and it held up for at least 2 more years, which is perfectly acceptable. So again, at the time they were right.

 

26 minutes ago, Stefan Payne said:

8 Cores at Desktop is 2 years old.

And we already have games that bitchslap 4 Core CPUs with an 8 Core one.

And that after ONLY 2 Years with available Hardware!

 

In a FEW games. Again, once the overall trend changes then so will I. 

 

27 minutes ago, Stefan Payne said:

Dude you realy have no idea what that means. And that you are totally wrong with your low core count CPU...

The only thing holding it back are the consoles right now. But this generation is on its way out, the next generation is already in development and possibly announced later this year or early next year!!

At that point, with a strong 8 Core probably 16 Thread CPU, there is no reason to hold back no more!!

 

ANd yes, Consoles are the primary development plattform right now...

Yeah, you're talking about next year for release, and then another year or 2 for developers to adjust to the new standard. In 3 years time people will be ready for a new CPU anyway.

 

29 minutes ago, Stefan Payne said:

No, its not.
Its lower than that.

ANd mostly only because of Clockrates.

the "IPC" you love so much, hardly increases.

THe last big step was Haswell.

I never said it was 7-8% in IPC. It's 7-8% in performance gains.

 

29 minutes ago, Stefan Payne said:

You are wrong.

As they only look at the higher end and rarely to never compare lower end.

And there the Intel don't look too good and are just too expensive.

No, they've looked at the lower end as well. Just look it up. They say the same thing. Jay even recommended a G4650 over a 220g.

[GoldenLag]

1 minute ago, jerubedo said:

they've looked at the lower end as well. Just look it up. They say the same thing. Jay even recommended a G4650 over a 220g

That is an old vid. Hardware unboxed does update and has said the new pentium is better than the 200GE. But only if you find it at MSRP, which you cant. 

 

And that was not even mentioning the upgradepaths. 

 

3 minutes ago, jerubedo said:

never said it was 7-8% in IPC. It's 7-8% in performance gains

Which are improvements in boost algorytm. Which is nice, but is honeslty performance which should been there earlier. 

4 minutes ago, jerubedo said:

That's a drop in the bucket compared to the thousands of games available on the market

Most of which are small budget indietitles which cant spend the extra on multicore workload sqedualling. 

[jerubedo]

33 minutes ago, GoldenLag said:

Depends entirely on budget. The budget space for the 2200G is tiny.  And the 8100 and 2200G doesnt occupy the same space. 

The 2200g is $95 and the i3 is $118. The i3 handily beats the 2200g in all games, so the extra ~$20 is well spent.

 

34 minutes ago, GoldenLag said:

We are allready seing a shift. If this was 2 years ago id agree with you, but there is allready a shift. Indietitles will continue to use singlecore or low core becuase its easier to develop. But its been that way since the first dualcore.

I would not consider the handful of games discussed a "shift." When the next 5 AAA games launch and they all take advantage of all cores/threads THEN I'll say there's a shift. 

 

37 minutes ago, GoldenLag said:

Depends entirely on the consumer. DDR5 will at the earliest get adopted in 2020. With 2021 which is what im looking at. 

 

PCIe standards will come if needed. Currently 4.0 is the new thing with 5.0 still being some time out. 

Yes but we're also talking about upgradeability. If 2021 comes around and the CPUs only support DDR5 then the upgrade path is still cut off for the consumer buying a board today, making it no better than having an intel setup on a dead-end socket.

 

40 minutes ago, GoldenLag said:

Last time i checked they gained nothing except increasing the boost clocks and pushing what can be called 95 watt TDP. 

That's correct but it still made all the difference.

 

41 minutes ago, GoldenLag said:

Good thing that is when AM4 ends and AM5 (presumably) starts.

Right, but all of you keep going on about how important an upgrade path is. If you're agreeing that most people will upgrade at 2021, then the upgrade path doesn't really matter does it?

[LogicWeasel]

5 minutes ago, jerubedo said:

You need to work on reading comprehension. I wasn't saying that Wifi and Bluetooth need extra bandwidth. I was talking about the extra 8 lanes that the PCIe-5.0 specification brings. Instead of 16 lanes minimum 5.0 defines a minimum amount of 24 lanes. Plugging in an M.2 drive and a wifi adapters brings your available lanes down to 8 for your GPU, which is enough for right now, but not the future. With 24 lanes we can have an M.2 drive, a wifi adapter, AND a GPU running at 16x. NVME drives will also get faster to take advantage of the extra bandwidth.

Did you really just try to insult him after failing to read your own sources, then suggest that a wifi adapter needs 4 or 8x lanes of PCIe 3.0 (let alone 5.0). 

 

Last I checked most wifi cards that use PCIe run on a 1x connector, but thanks for playing.  Also consumer M.2 drives use x4 lanes, so the math works out to more like x5 for lanes in-use, and it's worth noting that the PCIe 1x ports on most motherboards are directed through the express lanes provided by the motherboard, not the direct ones from the CPU.

[jerubedo]

31 minutes ago, Stefan Payne said:

Why not AMD?? 

What's wrong with AMD?

Again, Intel wins at every price point right now. The i5 is only $181.

 

32 minutes ago, Stefan Payne said:

If you recommend a  CPU/Plattform they can live with for 2 Years, you're giving bad advise.

If you recommend a CPU/Plattform they can live with for at least 4 Years, you don't give bad advise.

Simple as that.

First of all that's pretty arbitrary but ok. For a budget build you can't expect to keep excellent performance for 4 years. You can expect that from a mid-range to high-range build. But even at that, I think the i3 will still deliver the target 60 FPS in most games for 4 years. Time will tell. It's all speculation right now.

[jerubedo]

4 minutes ago, LogicWeasel said:

Did you really just try to insult him after failing to read your own sources, then suggest that a wifi adapter needs 4 or 8x lanes of PCIe 3.0 (let alone 5.0). 

 

Last I checked most wifi cards that use PCIe run on a 1x connector, but thanks for playing.  Also consumer M.2 drives use x4 lanes, so the math works out to more like x5 for lanes in-use, and it's worth noting that the PCIe 1x ports on most motherboards are directed through the express lanes provided by the motherboard, not the direct ones from the CPU.

No, when you plug in an M.2 wifi or Bluetooth adapter, you are using 4 lanes, not 1 lane. So using an M.2 drive and a wifi adapter will suck up 8 lanes. Not 5.

[jerubedo]

22 minutes ago, LogicWeasel said:

Intel favoritism

Again there's no favoritism here. I'm going with what's best. Once AMD wins I'll be recommending that and then arguing with the Intel people.