Why The Hate On AMD (Prosessors)

[thebennyboy]

This is nothing but fanboy vs fanboy idiocy that is not going to get anyone anywhere.

[patrickjp93]

Yes, I can. Because all that matters for gaming is cores, clockspeed, and gpu. Threads =/= Cores. Hyperthreading is a gimmick, prove to me that it's not. The clock efficiency and the performance are highers. MMOS? Died a long time ago. More taxing on the CPU? Games are games, they perform the same. Try crysis 3 on ultimate. Read my POSTS before. You aren't credible, deal with logic.

 

If an instruction takes an AMD chip 3 cycles to complete vs. 2 on an Intel chip, the AMD chip must run 50% higher clocks to have equal performance. Please get that simple fact through your skull!

 

Some instructions for rendering AMD can do in 2 cycles against Intel's 3, which is why the FX series is admittedly better for the task. This is also why Radeon cards are better for mining than Nvidia. There's one instruction rare to gaming but used heavily in scrypt mining/folding which Nvidia runs in 3 cycles to AMD's 1 cycle. That means the Nvidia card would have to have a 3x higher clock rate to be on par for that instruction.

 

This is what performance is. instructions per clock * clockspeed = throughput of 1 core. * core count = total performance.

 

Also, read the following technical documents on Hyperthreading. http://www.intel.com/content/www/us/en/architecture-and-technology/hyper-threading/hyper-threading-technology.html It's 2 logic cores per physical space.

 

a10 7850k has an IPC of .68 vs the I7 4790k IPC of .86 (consult the AMD and Intel x86 manuals for the Kaveri and Haswell architectures if you'd like to contest this). .68 instructions/cycle * 4.4 *10^9 cycles/second = 2,992,000,000 instructions per second on average assuming even distribution of instructions. I know this isn't the case as most people use only about 60% of the instructions on a CPU in daily tasks, but modeling this gets much harder.

.86 instructions/cycle * 4.4 * 10^9 cycles/second = 3,784,000,000 instructions per second. Now, the 4790k has double the core count of the a10 7850k until heterogeneous software comes along to use the graphics cores.

[zSanosake]

You're really upset. Tone it down then come at me with LOGIC. Hyperthreading is a gimmick. AMD is 2 cores per 4 module. That's stronger.

[Syntaxvgm]

You're really upset. Tone it down then come at me with LOGIC. Hyperthreading is a gimmick. AMD is 2 cores per 4 module. That's stronger.

Please please I want to hear you response to this. I'm taking you ignoring it as if you don't have one.

So so far, according to you,

most people here are wrong, everyone on OCN is wrong.

vr-zone is wrong

anandtech is wrong

hardwarepal is wrong

pcgameshardware is wrong

hardwarecanucks is wrong

pclab is wrong

and Passmark Software, makers of some of the most used benchamrking software is wrong?

They are ALL intel fanboys and this is all some conspiracy? You just said these, all in links we provided you, were fanboys?

and a video with linus explaining this is wrong?

[patrickjp93]

You're really upset. Tone it down then come at me with LOGIC. Hyperthreading is a gimmick. AMD is 2 cores per 4 module. That's stronger.

This is what performance is. instructions per clock * clockspeed = throughput of 1 core. * core count = total performance.

 

Also, read the following technical documents on Hyperthreading. http://www.intel.com...technology.html It's 2 logic cores per physical space.

 

A10 7850k has an IPC of .68 vs the I7 4790k IPC of .86 (consult the AMD and Intel x86 manuals for the Kaveri and Haswell architectures if you'd like to contest this). .68 instructions/cycle * 4.4 *10^9 cycles/second = 2,992,000,000 instructions per second on average assuming even distribution of instructions. I know this isn't the case as most people use only about 60% of the instructions on a CPU in daily tasks, but modeling this gets much harder.

.86 instructions/cycle * 4.4 * 10^9 cycles/second = 3,784,000,000 instructions per second. Now, the 4790k has double the core count of the a10 7850k until heterogeneous software comes along to use the graphics cores too, so the idea the A10 7850k = 4790k currently is an egregious lie.

 

And this was using the 7850k Black Edition, which is clocked .3GHz higher than the basic version.

[Wano97]

Is this still going on  :rolleyes: ? He's obviously repeating the same shit all over and doesn't even say anything useful besides 'AMD is superior'

Tl:dr it's a troll

[Syntaxvgm]

Ladies, please lower your handbags. This is not getting anywhere, surprised a mod hasn't closed this thread already.

This doens't involve you. I bet you are willing to learn and if someone showed you almost literally every source on the internet for a tech fact. You wouldn't deny it without proving any source of your own.

That's the problem. I learn new stuff all of the time. Everyone does. He refuses.

edit lol mod is really active on this thread. His post is gone

[patrickjp93]

This doens't involve you. I bet you are willing to learn and if someone showed you almost literally every source on the internet for a tech fact you wouldn't deny it without providng any source of your own.

That's the problem. I learn new stuff all of the time. Everyone does. He refuses.

When we refuse to accept we are limited in our understanding of the world, we only serve to stifle it further. I like being PROVEN wrong, because then I learn something new and become more learned. This is why U.S. schooling is going down the drain: no willingness to be wrong. Emphasizing perfection is the surest path to the destruction of learning.

[thebennyboy]

This doens't involve you. I bet you are willing to learn and if someone showed you almost literally every source on the internet for a tech fact you wouldn't deny it without providng any source of your own.

That's the problem. I learn new stuff all of the time. He refuses.

 

Whether he is right or wrong means nothing, the fact you're just adding fuel to the fire by arguing back doesn't stand in your defence. You made your point and showed the sources, if that isn't enough then just leave it. Either way, this thread is getting nowhere are should be closed.

[Syntaxvgm]

Whether he is right or wrong means nothing, the fact you're just adding fuel to the fire by arguing back doesn't stand in your defence. Either way, this thread is getting nowhere are should be closed.

Nah I need to vent. I never argue on the internet. Let me have fun.

I mean, not that you are wrong or anything.

[patrickjp93]

I mean, not that you are wrong or anything.

Nice save  :rolleyes:

[Syntaxvgm]

When we refuse to accept we are limited in our understanding of the world, we only serve to stifle it further. I like being PROVEN wrong, because then I learn something new and become more learned. This is why U.S. schooling is going down the drain: no willingness to be wrong. Emphasizing perfection is the surest path to the destruction of learning.

Omg yes. The thing he believes I think all new techies believe for a minute, it's direct number comparison that would seem to make sense to someone new. But we all learn and move on. I have learn so much over the years from other to kindly point out when I'm wrong, or answer questions I had.

[thebennyboy]

At this point you are doing nothing but feeding the trolls. You've added your arguements and evidenced it, if he doesn't like it then just leave it.

[Syntaxvgm]

Nice save  :rolleyes:

Hes right. All I'm accomplishing is nothing but venting.

[patrickjp93]

But at the end of the day AMD has better pricing for low to even mid-range chips and has some surprisingly good options for video rendering. Its Heterogeneous Systems Architecture is revolutionary and poses enough of a long-term threat Intel is copying the move in time to catch the storm of major software makers implementing it. At that point both companies' APUs will be far better than what we have now, and the dynamics of supremacy may change drastically, but for now, AMD=price/performance and Intel = raw performance as well as some very nifty features for distributed systems like trusted execution and vPro tech for cyber defense (if AMD has equivalent features somewhere I've never heard of or seen mention of them).

[dizmo]

Because 95% of people run at stock speeds!

 

I personally don't, but most people stay at the recommended speed since upgrades happen on a 4-year cycle for major enthusiasts and on 8-year cycles for people who do nothing but browse, do office work, and watch movies.

Haha, that is entirely me. I always get the unlocked processors and Z series boards and...leave them at stock clock.

I upgrade every 4 years as well. Whenever I get bored with what I have and the buy itch comes. Next year is new PC year

 

 

so is it worth it to get a 4960X over a 4930K because i'm probably gonna get a 4930K when i build a new PC

No, over and over again no.

A 4930k might even be unnecessary.

[patrickjp93]

Maybe I'm inviting disaster, but it would seem our die-hard AMD fanboy has either run off or is actually reading the Intel Docs I posted. Let's hope for the latter.

[FlamingMetalHorse]

I was about to respond to that one guy (you know who I'm talking about) about how he is wrong and how he should reevaluate the facts, or at lease consider them, but in the past three hours that I prepared my six paragraph retort, you all did it for me, and I thank you all.

 

I don't hate AMD (I actually love AMD), but I would just like to see them come out with an extreme enthusiast grade CPU that would compete with Intel's extreme enthusiast grade processors (Ie: Ivy Bridge-E, upcoming Haswell-E, you get it).

[Syntaxvgm]

I was about to respond to that one guy (you know who I'm talking about) about how he is wrong and how he should reevaluate the facts, or at lease consider them, but in the past three hours that I prepared my six paragraph retort, you all did it for me, and I thank you all.

 

I don't hate AMD (I actually love AMD), but I would just like to see them come out with an extreme enthusiast grade CPU that would compete with Intel's extreme enthusiast grade processors (Ie: Ivy Bridge-E, upcoming Haswell-E, you get it).

I'm more concerned about the mid range and efficiency in order to make a better high end. The 220w 9590 @ 4.7-5.0ghz matches the 4770k in some measurements, but that's the top-top of their line and produces more heat than a quasar. I'd be a little happier if it had more OC room.

[Sintezza]

facepalm me all you want. These are facts. 

You fall into the group that thinks more cores and higher clockrate always equals better performance. That is simply not true. 

This is not fanboyism. I would prefer AMD to have a better lineup over intel. 

 

LOL dude

 

you are the one who makes false statements here.

you say that i belong to a specific group without even knowing me.

 

I dont like that, you definitely dont have to teach me anything about computer hardware in general.

If you think that i am one of those people who belongs in that group, then you are stupid lol.

 

"" cores and higher clockrate always equals better performance.""

 

I never made that statement, its something you fantasize your self.

 

you are just annoying nothing more then that.

Oh gosh i let me tricked into this bull shit again.

okay im done!

[Faa]

but I would just like to see them come out with an extreme enthusiast grade CPU that would compete with Intel's extreme enthusiast grade processors (Ie: Ivy Bridge-E, upcoming Haswell-E, you get it).

That isnt possible for them atm.

1) Power consumption would be a big problem; if they add another series of 4 PD modules that would kick them at 300-400W load who knows. Shipping such a CPU with a 2.0GHz clock to counter the power consumption, not many people will be interested in it except for people who are purely after multithreaded performance and give a jack about single threaded/gaming performance since that thing would perform worse than a regular vishera.

2) Most boards wont be able to feed such a cpu, the boards would be extremely pricey.

3) Socket needs a swap, plenty of people reported that their socket melted.

4) In terms of performance; IPC is just too low to compete with SB/E IB/e or even the upcoming octacore.

Their best bet currently is; bumping their IPC massively and not just 40% (preferably above Haswell) so they dont have to use ridiculous amount of cores to lower the power consumption further down. In the end CPU monopoly purely comes down to IPC, pair it with low power consumption youre the leader. Afaik AMD made recently a deal with GF for 14nm, dropping CMT (module nonsense) and continuing their FX line.

 

 

Now, here's where we have to draw the line between objective views and fanboyism. The claim that AMD is cooler-running is simply not true. 32nm process is only partly to blame. AMD is quite behind the curve in terms of power consumption and heat; I'm not exactly sure what causes this other than 32nm, because Steamroller is a 28nm part and it still draws considerable power compared to Haswell quad-cores. 

A clear example could be the 480 vs 580, same process, almost the same die size but the 580 runs noticeably cooler. They improved the transistors if I'm not mistaken to achieve this. That AMD is quite behind Intel in terms of power consumption? Havent checked their steamrollers yet, but a 3930K is more of a fair comparison to put against a 8350 although Im pretty sure if the 3930K was an octacore it definitely would pass a 8350 but yes their performance/wattage is awful atm.

 

 

 AMD also makes really bad temp sensors in their CPUs, which causes absurd idle temperature readings such as 8°C under a AIO cooler, which isn't physically possible unless you live in an extremely cold environment. 

They dont make bad temp sensors but theyre being stupid. The temp reading what you see in your monitoring tools arent the raw data from the temp sensor like on Intels, theyre just using some algorithm predictions to measure the cpu temp instead of providing the actual data. The hotter the cpu is, the more accurate the reading becomes >.< Official amd response regarding this: http://forums.amd.com/game/messageview.cfm?catid=446&threadid=160921&enterthread=y

So you can't really compare Intel temperatures with AMD. Measuring it from the heatspreader? Intel cpu's barely bump heat transfer heat to the HSP, die vs die comparison seems to be the most accurate but gl taking a soldered HSP off without damaging the die.

[FlamingMetalHorse]

 

That isnt possible for them atm.

1) Power consumption would be a big problem; if they add another series of 4 PD modules that would kick them at 300-400W load who knows. Shipping such a CPU with a 2.0GHz clock to counter the power consumption, not many people will be interested in it except for people who are purely after multithreaded performance and give a jack about single threaded/gaming performance since that thing would perform worse than a regular vishera.

2) Most boards wont be able to feed such a cpu, the boards would be extremely pricey.

3) Socket needs a swap, plenty of people reported that their socket melted.

4) In terms of performance; IPC is just too low to compete with SB/E IB/e or even the upcoming octacore.

Their best bet currently is; bumping their IPC massively and not just 40% (preferably above Haswell) so they dont have to use ridiculous amount of cores to lower the power consumption further down. In the end CPU monopoly purely comes down to IPC, pair it with low power consumption youre the leader. Afaik AMD made recently a deal with GF for 14nm, dropping CMT (module nonsense) and continuing their FX line.

 

Holy crap, not with Vishera! Of course they're going to need a new socket and CPU architecture! I probably should have specified that.

[Guest]

Holy crap, not with Vishera! Of course they're going to need a new socket and CPU architecture! I probably should have specified that.

 

FX-9590 was AMD's attempt at a IVB-E competitor.

 

They couldn't pull it off because an AIO was mandatory at the very least to cool it,

 

Half of the AM3+ boards aren't even 1/4 decent, so AMD had to warn against using it on anything other than top of the line 990fx,

 

And it wasn't really any different from an OC 8350. They really have to ditch CMT before they can take on X99, and they better do it before well into 2015 like Amd plans.

 

Furthermore, the voltage required to push 4.5ghz on a 8350 is rather unacceptable by today's standards.

[Rubenis]

My one Euro about this is as follow:

 

Low on cash - buy entry level AMD, like Athlon X4 750K.

Have normal amount of cash - i5 is the way.

Have a lot of cash - Again Intel, but depending on needs.

 

I have Intel just because of one reason - efficency. If Intel 4 core i5 4670K is better 9 times out of 10 against FX 8350 8 core processor with twice as less cores and lower core clock and lower TDP it is hell of a lot more efficent.

 

AMD strategy is quite clear - people with low confidence (smart people on budget use this situation). Because it is like shouting - 8 cores, 5Ghz processor, beast rrrrr. Actually people end up with 220W TDP and single threaded performance, that even i3 can beat.

[Bloodkil933]

Threads like this are really starting to piss me off, people still hate Vishera for some unknown reason, Understandably it has a lower IPC than Intel, but as Logan showed, the 8320/8350's were and still are VERY good value for money.

 

Any day of the week i'd recommend an AMD 8320 to someone wanting to game and to just overclock it. And before anyone brands me as a fanboy which seems to be a common thing at the moment, the Intel CPU's are better, but I would argue not for gaming, any game which can fully utilise those 8 cores will see fairly similar performance to an Intel machine.