a look at (ry)Zen's IPC - from Excavator to KabyLake

[79wjd]

8 minutes ago, Liltrekkie said:

"What appears to be the limit of the chip" Well I'm so sorry to those people who got the short end of the silicon lottery. It was known that the 1700x and 1800x are binned higher than the 1700. 

 

So.. AMD told reviewers to run memory in duel channel mode? Oh god the horror. What a crime my god AMD needs to burn for that! And the horror for wanting to use other benchmarks to show that a game wouldn't be CPU bottlenecked. Oh god how dare AMD do this!!

 

I'm so glad you have all this inside information about AMD giving mobo makers 3 weeks for a BIOS, even though MOBO leaks have been happening for the better part of 2 months. 

 

AMD didn't have a perfect launch? No one does. But lets burn them to the ground because they are AMD. 

 

Intel: skylake bendgate. skylake supply issues. skylake bug causes application freeze during complex workloads, haswell has USB 3 issues (fixed though) haswell overheating issues. poor TIM. 

 

Both sides have had issues. 

You mean, like reviewers who always get the creme de la creme? If their chips can't overclock, then it's pretty safe to say that the average chip won't overclock well either. Even the 1800x that reviewers got aren't really going past 4/4.1ghz (the same as the 1700's).

 

Actually yes....that's very bad. First off, quad channel memory does make a difference in certain tasks. Secondly, MCE also makes a big difference in tasks (but it's okay to let AMD's XFR run....gotta make it far afterall). Thirdly, you WANT the CPU to be the bottleneck when you're comparing CPUs -- that's why CPU performance tests are usually done at <=720p -- those tests show the relative performance between CPUs. Benchmarking a CPU at 4k where the GPU is the limiting factor doesn't show anything about how CPU performance compares. And sure, someone who's going to buy a 1800x for gaming probably won't be running it at 1080p, but what happens in three years when GPUs are far more powerful and the CPU is now going to be the limiting factor? The 4k benchmark then will show the same thing that the 1080p benchmarks show today. So yes, it is a big problem. 

 

They have had information about what kind of features the board would need for more than three weeks, but they haven't had specific information required to create a bios for Ryzen.

 

I never said Intel hasn't had issues, X99 was also a mess at launch. It's never one factor that determines whether a launch is good or bad. It's everything put together, and the Ryzen launch has had many issues -- from both a technical and marketing standpoint that have led to the Ryzen launch being a bad one.

[Liltrekkie]

1 minute ago, djdwosk97 said:

You mean, like reviewers who always get the creme de la creme? If their chips can't overclock, then it's pretty safe to say that the average chip won't overclock well either. Even the 1800x that reviewers got aren't really going past 4/4.1ghz (the same as the 1700's).

 

Actually yes....that's very bad. First off, quad channel memory does make a difference in certain tasks. Secondly, MCE also makes a big difference in tasks (but it's okay to let AMD's XFR run....gotta make it far afterall). Thirdly, you WANT the CPU to be the bottleneck when you're comparing CPUs -- that's why CPU performance tests are usually done at <=720p -- those tests show the relative performance between CPUs. Benchmarking a CPU at 4k where the GPU is the limiting factor doesn't show anything about how CPU performance compares. And sure, someone who's going to buy a 1800x for gaming probably won't be running it at 1080p, but what happens in three years when GPUs are far more powerful and the CPU is now going to be the limiting factor? The 4k benchmark then will show the same thing that the 1080p benchmarks show today. So yes, it is a big problem. 

 

They have had information about what kind of features the board would need for more than three weeks, but they haven't had specific information required to create a bios for Ryzen.

 

I never said Intel doesn't have issues either, X99 was also a mess at launch. It's never one factor that determines whether a launch is good or bad. It's everything put together, and the Ryzen launch has had many issues -- from a technical and marketing standpoint that have led to the Ryzen launch being a bad one.

 

So... You're mad because AMD didn't make a direct competitor for intels HEDT line that has quad channel memory? Wtf is wrong with you? And sorry, more gamers are becoming CPU bound by the day, so yes it actually makes SENSE to push for higher resolutions as they become mainstream. 1080p is basically the new 720p when compared to 1440p. I'm pretty sure AMD wanted to look good for those people who game and do stuff at those resolutions. 

 

And if they have had such issues, how come Ryzen was demo'd so much? There were tons of demo systems around. They all obviously had a BIOS. 

[GDRRiley]

1 hour ago, KOMTechAndGaming said:

but ddr4-2400mhz should be fine right?

yeah

[79wjd]

20 minutes ago, Liltrekkie said:

 

So... You're mad because AMD didn't make a direct competitor for intels HEDT line that has quad channel memory? Wtf is wrong with you? And sorry, more gamers are becoming CPU bound by the day, so yes it actually makes SENSE to push for higher resolutions as they become mainstream. 1080p is basically the new 720p when compared to 1440p. I'm pretty sure AMD wanted to look good for those people who game and do stuff at those resolutions. 

 

And if they have had such issues, how come Ryzen was demo'd so much? There were tons of demo systems around. They all obviously had a BIOS. 

No. The fact that Ryzen doesn't have quad channel memory is fine given that it is much cheaper. The problem is that AMD asked reviewers to pretend like X99 doesn't have quad channel memory in order to make the Intel results look worse and thus make the AMD results look better. No one would run an X99 CPU with dual channel memory, therefore asking reviewers to gimp their X99 CPUs is clearly only done to make AMD look better in comparison than it actually is. 

 

Yes, higher resolutions are more relevant in the category that an 1800x fits into. HOWEVER, in three years time the SAME bottlenecks that you see now at 1080p will be occurring at 4k. So while right now, at 4k, Ryzen performs close to Intel's offerings, in the future, as GPUs get faster, the performance delta will increase. The 1080p (and lower) results today are indicative of what future 4k results will be. Right now, at 4k, games are mostly GPU limited rather than CPU limited, but as GPUs get faster we will reach a point where the GPU is more than fast enough to keep up with CPUs like a 1800x, thereby causing a CPU bottleneck (one that wouldn't occur with a 7700k until GPUs got even faster). That is why CPU testing (for gaming performance) is done at low resolutions -- the goal is to cause a CPU bottleneck so you can compare the absolute performance of CPUs when they're not being hindered by other components (e.g. the GPU). 

 

For all you know those demo systems were using an AMD designed/built board and/or with an AMD bios. I can promise you that AMD has been testing physical Ryzen chips long before they gave the specs to board manufacturers. 

[straight_stewie]

6 hours ago, zMeul said:

source: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/

 

"The Slit", established member over Anand's forums has done some testing that feature more technical benchmarks that you don't usually see

go check the source for all the benches and results plus in depth technical details on the CPU and chipset (Zeppelin)

 

the tests were conducted at 3.5Ghz

mobo used ASUS CROSSHAIR VI HERO

results with ER mean extremes removed (highest and lowest)

 

let's talk IPC and how it compares to CPUs ranging from Excavator era to Intel's latest KabyLake:

These are much more fair benchmarks. And I would have to say that I agree with your general synopsis that it would seem as if Ryzen is about half as fast as comparably complex Intel chips.

So I will just say three things: You have partially changed my opinion, Ryzen costs half as much as comparably complex Intel processors (and receives half the performance), And a quote from "The Slit":

Quote

850 points in Cinebench 15 at 30W is quite telling. Or not telling, but absolutely massive. Zeppelin can reach absolutely monstrous and unseen levels of efficiency, as long as it operates within its ideal frequency range.

The i7-4770k produced the same score at 120W. 

As such, while Ryzen may not be what it was hyped up to be, but it would still make an excellent laptop processor, providing more than good enough performance to play AAA games, at rather low power consumption.

According to The Slits graph, the peak Cinebench R15 score for the Ryzen 1800x was 1450 at 75 Watts. This puts it in the range of Intels E5-2630 v2 chip, which produces the same score at $80 watts and steady prices of more than $650. 

Drawing from those two conclusions, I will make this final statement: You are correct, Ryzen is not the end all be all (or even near the best) gaming chip. The Ryzen 1800x, however, seems like it would make a good all around dual workstation/gaming laptop rig, or even a decent dual workstation/gaming desktop. 

[zMeul]

4 minutes ago, straight_stewie said:

The i7-4770k produced the same score at 120W. 

As such, while Ryzen may not be what it was hyped up to be, but it would still make an excellent laptop processor, providing more than good enough performance to play AAA games, at rather low power consumption.

but you're comparing it with a 4y old CPU, what about mobile Kabys?

 

the CineBench scores .. CineBench also does OpenGL rendering GPU accelerated

why would you buy a expensive CPU when you can buy a less expensive one, pair it with a decent video card and achieve much more performance at the same costs?

 

also, the mobile parts are not out .. yet

the mobile parts will also have iGPUs attached to them, increasing the power requirements

[straight_stewie]

19 minutes ago, zMeul said:

also, the mobile parts are not out .. yet

the mobile parts will also have iGPUs attached to them, increasing the power requirements

It is possible to use desktop grade components in a mobile environment you know (Linus proving this). You want to compare it to a Kaby Lake chip, fine: Intels i7-7700k gets a Cinebench score of 1500 at $350 and 91 watts. 

The numbers I'm looking at came from a test where the same GPU and memory was used.

So my point still stands. The Ryzen is a mobile capable i7-7700k, in desktop form. And just to be clear, you can easily make a low power desktop chip a mobile chip simply by switching the socket PCB out for a solder-in PCB.

So please, tell me again how Ryzen sucks.

 

[zMeul]

30 minutes ago, straight_stewie said:

It is possible to use desktop grade components in a mobile environment you know (Linus proving this). You want to compare it to a Kaby Lake chip, fine: Intels i7-7700k gets a Cinebench score of 1500 at $350 and 91 watts. 

The numbers I'm looking at came from a test where the same GPU and memory was used.

So my point still stands. The Ryzen is a mobile capable i7-7700k, in desktop form. And just to be clear, you can easily make a low power desktop chip a mobile chip simply by switching the socket PCB out for a solder-in PCB.

So please, tell me again how Ryzen sucks.

you if people are willing to buy that monstrosity they need to check their sanity at the door - those "laptops" are what? over 2grand .. why?!?!

 

Edited March 4, 2017 by zMeul

[straight_stewie]

12 minutes ago, zMeul said:

you people are willing to buy that monstrosity they need to check their sanity at the door - those "laptops" are what? over 2grand .. why?!?!

 

I'm not willing to buy anything. I have an i7-4770k and twin 760TIs. I'm still playing AAA games at high or higher settings at above 60 FPS. I'm perfectly happy.

I'm saying that you've set out on this multithreaded rampage to change peoples opinion and you aren't willing to evaluate the validity of your own in the slightest, even when presented with information from an article that you quoted that shows something contradictory to your blanket opinion about Ryzen. 

Ryzen has it's place and it's exactly what I said it was: It would be great paired with a GTX1070M in a 17 inch ~$1000 US laptop. There are a couple of other similar configurations I could think of as well that would suit the chip. That's the market for the Ryzen chips, the only remaining question or doubt is "If/when will AMD figure this out?"

[zMeul]

12 minutes ago, straight_stewie said:

I'm not willing to buy anything.

sorry, not "you people" 

"if people"

 

I edited my previous post

 

Quote

Ryzen has it's place and it's exactly what I said it was: It would be great paired with a GTX1070M in a 17 inch ~$1000 US laptop. There are a couple of other similar configurations I could think of as well that would suit the chip. That's the market for the Ryzen chips, the only remaining question or doubt is "If/when will AMD figure this out?"

yes, but you are talking about something that does not exist .. yet

when that something materializes it, then we'll have a new conversation

the discussion at hand is with the products that exist

 

if you want to talk ifs and whens, why wouldn't Intel release new SKUs both desktop and mobile to directly attack Ryzen's price point - because right now, Intel does not have products in that spot

[Lays]

33 minutes ago, straight_stewie said:

It is possible to use desktop grade components in a mobile environment you know (Linus proving this). You want to compare it to a Kaby Lake chip, fine: Intels i7-7700k gets a Cinebench score of 1500 at $350 and 91 watts. 

The numbers I'm looking at came from a test where the same GPU and memory was used.

So my point still stands. The Ryzen is a mobile capable i7-7700k, in desktop form. And just to be clear, you can easily make a low power desktop chip a mobile chip simply by switching the socket PCB out for a solder-in PCB.

So please, tell me again how Ryzen sucks.

 

7700k ain't gettin 1500 R15 unless it's on Ln2 at near 7 ghz lol

[straight_stewie]

3 minutes ago, Lays said:

7700k ain't gettin 1500 R15 unless it's on Ln2 at near 7 ghz lol

https://www.extremetech.com/extreme/241950-intels-core-i7-7700k-reviewed-kaby-lake-debuts-desktop

I think you may be right. I may have given the Intel chip *way too much credit.

[Lays]

11 minutes ago, straight_stewie said:

https://www.extremetech.com/extreme/241950-intels-core-i7-7700k-reviewed-kaby-lake-debuts-desktop

I think you may be right. I may have given the Intel chip *way too much credit.

I don't even think I get 1200 at 5.2g, so yea you gave it a bit too much credit lol

[PCGuy_5960]

7 hours ago, GDRRiley said:

it was just stated that the ryzen mobo used has isssue and that it isn't the full story nor will it be for weeks. 

The MSI Mobo had issues that affected performance.... The ASUS one didn't have such issues. Nor did the Gigabyte and AsRock motherboards, and tests using these boards have also shown that Ryzen's IPC is 5% lower than Broadwell-E, so slightly lower than Haswell...

[RagnarokDel]

9 hours ago, zMeul said:

and that's AMD's / ASUS' problem

 

when they fix their shit, maybe he'll retest

Yeah let's make a claim for absolute results. Imma use a the motherboard I know is known to be problematic, there's no chance anyone would cry foul, is there? You're so intellectually dishonest, it's baffling.

[Godlygamer23]

9 hours ago, TOMPPIX said:

do you see it drop under 60 frames?

This is a lame argument. I game ideally at 144FPS. I don't even want my games dipping below 144FPS.

[Glenwing]

10 hours ago, ace_cheaply said:

Correct me if I'm wrong please, but is this saying that kabylake has 25-30% ipc advantage over haswell?  I don't believe that to be the case.

 

10 hours ago, djdwosk97 said:

According to this, it shouldn't be, but reviews put it around Broadwell-E (outside of gaming). Not to mention these results put Kaby Lake significantly further ahead of Haswell than it should be. 

No, the scores here show Kaby Lake to be about 17.3% higher than Haswell. It's just because they're percentages, not absolute numbers, and percentages depend on reference point.

 

For example:

 

A = 50

B = 100

C = 150

 

If we compare these as a percentage of A:

 

A = 100% of A

B = 200% of A

C = 300% of A

 

Notice that C is 100 percentage points above B when compared like this, even though C (150) is only 50% higher than B (100).

 

In this case, yes Kaby Lake is 180% and Haswell is 150%. However, both of these percentages are "percent of Excavator IPC". So Kaby Lake is 30%-of-Excavator's-IPC above Haswell. But when people ask what the IPC of Kaby Lake is compared to Haswell, usually they want to know how many Haswells-worth of IPC higher it is, not how many Excavators-worth of IPC it is. For that, you would need to convert from the numbers in this chart (just divide the Kaby Lake percentage by the Haswell percentage to get the percentage of Kaby Lake above Haswell).

[tlink]

9 hours ago, zMeul said:

there is nowhere to blossom

with CPUs, what you see is what you get

 

it's not like you release a new graphics driver and get couple of % perf gains in some titles

yes, AMD is capable of releasing micro-code updates to fix some of the shit, but don't expect to magically close the gap

 

AMD needs to get their asses in gear right now for Zen2, if they aren't already on it

I agree somewhat, but at the end of the day we simply can't know what the future might bring to it. I'm for example really interested into what them working with devs for optimization will result into.

[tlink]

44 minutes ago, Glenwing said:

 

No, the scores here show Kaby Lake to be about 17.3% higher than Haswell. It's just because they're percentages, not absolute numbers, and percentages depend on reference point.

 

For example:

 

A = 50

B = 100

C = 150

 

If we compare these as a percentage of A:

 

A = 100% of A

B = 200% of A

C = 300% of A

 

Notice that C is 100 percentage points above B when compared like this, even though C (150) is only 50% higher than B (100).

 

In this case, yes Kaby Lake is 180% and Haswell is 150%. However, both of these percentages are "percent of Excavator IPC". So Kaby Lake is 30%-of-Excavator's-IPC above Haswell. But when people ask what the IPC of Kaby Lake is compared to Haswell, usually they want to know how many Haswells-worth of IPC higher it is, not how many Excavators-worth of IPC it is. For that, you would need to convert from the numbers in this chart (just divide the Kaby Lake percentage by the Haswell percentage to get the percentage of Kaby Lake above Haswell).

This always is a tricky thing because data visualization isnt a very exact science, yet a small difference can make a huge impact in the impression it will give. this is why I prefer raw data over statistics.

[Misanthrope]

8 hours ago, Liltrekkie said:

 

So... You're mad because AMD didn't make a direct competitor for intels HEDT line that has quad channel memory? Wtf is wrong with you?

Yeah wtf is wrong with you!!?? Why would you expect AMD to at lest come close to intel in IPC? JUST SHUT UP AND WAIT ANOTHER 5 YEARS!!!

[RotoCoreOne]

So basicly, if you're aiming for absolute gaming performance get an i7 7700k. If you're looking at Broadwell-E for workstation purposes but don't want to burn cash go with ryzen

[Kinda Bottlenecked]

 

Basically 

 

Win 7 lel

[zMeul]

1 hour ago, tlink said:

I agree somewhat, but at the end of the day we simply can't know what the future might bring to it. I'm for example really interested into what them working with devs for optimization will result into.

then you'd have Zen exes and rest of the world exes

and then you'll have people who'll accuse AMD for specifically compiling for them - like they're now accusing Intel of 

[laminutederire]

2 hours ago, Godlygamer23 said:

This is a lame argument. I game ideally at 144FPS. I don't even want my games dipping below 144FPS.

That's an as lame argument since you're just extremely exigent. Loads of people are fine with 70-80 fps steady, why do you guys have to shit all over products which are perfectly fine for many people who would be quite okay with it.

It's not the ultimate gaming chip, but it's decent enough. As some reviewers point out, it's on par or better than i7 for many workstation workloads, and is on par with i5 and could maybe be better when amd fixes issues.

Since when i5 performance is bad? It's not optimal but it is in no case bad.

You have those 144 fps in e sports titles, and you have 70+ fps in all the other titles, so it's quite alright to be fair for someone who need the work while not having a second rig to game on, while still not paying twice the price.

[zMeul]

9 minutes ago, laminutederire said:

It's not the ultimate gaming chip, but it's decent enough. As some reviewers point out, it's on par or better than i7 for many workstation workloads, and is on par with i5 and could maybe be better when amd fixes issues.

here's the thing tho, the i7s are cheaper (up to a point)

and the i5s are cheaper still

 

why would you pay premium for less performance? it doesn't make any sense