Page 1 of 3
1
2
3
LastLast
  1. #1

    AMD vs Intel for gaming article

    http://www.technologyx.com/featured/...ce-showdown/4/

    I've been researching a new build based around dx12. Mostly looking at high-end AMD processors vs comparably priced Intel. I found this article and the results seem to be rather interesting. AMD on dx12 seems to be getting up there.

    I'm not in one camp or the other when it comes to processors, although I have had Intel for the past 7 years. After reading this, I'm not sure I would dismiss AMD for basic gaming. My biggest concerns would be power draw. The rated power on the AMD is like 125w were the i7 is like 85ish watts.

    At idle and basic computing(non gaming) usage, does power consumption matter? Or is it more of a max load type thing?

    Can we have a discussion on dx12 amd vs intel logically without a heavy dose of bias on this forum?

  2. #2
    Quote Originally Posted by Lightwysh View Post
    Can we have a discussion on dx12 amd vs intel logically without a heavy dose of bias on this forum?
    Probably not.

    I'm not entirely sure what to read out of the article really, or to be more precise what to read as the cause behind the performance, seeing as the results are somewhat inexplicable and weird. I mean, this is the benchmarks of each of them compared to each other on anandtech.com... (Unless I found the wrong AMD processor, which is possible.)

  3. #3
    Quote Originally Posted by Drunkenvalley View Post
    Probably not.

    I'm not entirely sure what to read out of the article really, or to be more precise what to read as the cause behind the performance, seeing as the results are somewhat inexplicable and weird. I mean, this is the benchmarks of each of them compared to each other on anandtech.com... (Unless I found the wrong AMD processor, which is possible.)
    Your article shows a wider variance in FPS and the only difference I see is, your article has a 770 where the one I linked has a 970. I also think the one I linked is dx12 based, the Anandtech article does not mention dx12(I may not be seeing it).

    Could the GPU make that big of an impact?

  4. #4
    That article is terrible. It's comparing CPU's with highly graphically bound workloads - not useful at all.

  5. #5
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Drunkenvalley View Post
    Probably not.

    I'm not entirely sure what to read out of the article really, or to be more precise what to read as the cause behind the performance, seeing as the results are somewhat inexplicable and weird. I mean, this is the benchmarks of each of them compared to each other on anandtech.com... (Unless I found the wrong AMD processor, which is possible.)
    You found the wrong CPU, the E variant is clocked lower for lower TDP. (3.3GHz vs. 4.0GHz)

    Regardless the point of his thread was gaming performance where Anand Tech's bench is synthetic/graphics/video benchmarking with a little bit of gaming.
    As the benchmark showed previously it does take a bit of overclocking to get the AMD 8-core up to snuff and running at 4GHz makes a difference.

    There's no point to compare AMD 8-core vs. Intel 8-core with HT, AMD doesn't stand a chance in the expensive suite.
    But than the AMD CPU is built to go up against Intel's Quad Core i5/i7 range and it is beaten but not by a large margin.

    Only in Single threaded does it get thrashed vs. them but any game that is optimised will get a run for it's money when it's AMD vs. Intel really.

    Some people just think AMD's CPUs are entirely garbage which is one of the biggest loads of shit there is.

    Are they older and "less powerful" than Intel CPUs? Yes.
    Does that mean they perform any worse? Not necessarily.
    Last edited by Evildeffy; 2015-09-03 at 01:54 PM.

  6. #6
    Quote Originally Posted by Evildeffy View Post
    You found the wrong CPU, the E variant is clocked lower for lower TDP. (3.3GHz vs. 4.0GHz)

    Regardless the point of his thread was gaming performance where Anand Tech's bench is synthetic/graphics/video benchmarking with a little bit of gaming.
    As the benchmark showed previously it does take a bit of overclocking to get the AMD 8-core up to snuff and running at 4GHz makes a difference.

    There's no point to compare AMD 8-core vs. Intel 8-core with HT, AMD doesn't stand a chance in the expensive suite.
    But than the AMD CPU is built to go up against Intel's Quad Core i5/i7 range and it is beaten but not by a large margin.

    Only in Single threaded does it get thrashed vs. them but any game that is optimised will get a run for it's money when it's AMD vs. Intel really.

    Some people just think AMD's CPUs are entirely garbage which is one of the biggest loads of shit there is.

    Are they older and "less powerful" than Intel CPUs? Yes.
    Does that mean they perform any worse? Not necessarily.
    Isnt this what Win10 and dx12 are hoping to accomplish? Make more usage out of multi-core processors? Better performance from more cores?

  7. #7
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Lightwysh View Post
    Isnt this what Win10 and dx12 are hoping to accomplish? Make more usage out of multi-core processors? Better performance from more cores?
    Fundamental difference there really.
    The workload gets spread and worked with across multiple/all cores.

    The way you're saying it is that it is done to promote performance first, the rest later.
    In fact the reason it's done is to promote technological advancement with a result that performance increases.

    We've hit a "physical" limit in regard to clock speeds, AMD knew this weakness and advanced that technology with multi-cores and increased IPC.
    This was when the first AMD64 CPUs were born which annihilated Intel's Pentium 4/D series.
    Hell if you look at the patents and licenses it is Intel which pays AMD an X amount to allow the use of Multi-core technology/Integrated Memory Controller etc.

    AMD then went lax about it and Intel caught up and the repercussions are felt still because of that.

    It's true that AMD did accelerate this by creating Mantle, which has done it's job splendidly, because there's no reason to progress if people are paying shittons of money for things that work fine as they are.

    In truth our parallelism technology is far ahead of what we're using simply because of the lack of "need" to develop it because of money.

    Imagine the following:
    Had AMD not pushed Mantle as far as they did giving developers exactly what they've been screaming for for years, would you have DX12 right now?
    Very likely you would, but you would not have control over the resources of the GPU as much as you would have now.

    In fact I am pretty sure that DX12 would've been far less control over the GPU's resources than it is now.

  8. #8
    Quote Originally Posted by Evildeffy View Post
    Fundamental difference there really.
    The workload gets spread and worked with across multiple/all cores.

    The way you're saying it is that it is done to promote performance first, the rest later.
    In fact the reason it's done is to promote technological advancement with a result that performance increases.

    We've hit a "physical" limit in regard to clock speeds, AMD knew this weakness and advanced that technology with multi-cores and increased IPC.
    This was when the first AMD64 CPUs were born which annihilated Intel's Pentium 4/D series.
    Hell if you look at the patents and licenses it is Intel which pays AMD an X amount to allow the use of Multi-core technology/Integrated Memory Controller etc.

    AMD then went lax about it and Intel caught up and the repercussions are felt still because of that.

    It's true that AMD did accelerate this by creating Mantle, which has done it's job splendidly, because there's no reason to progress if people are paying shittons of money for things that work fine as they are.

    In truth our parallelism technology is far ahead of what we're using simply because of the lack of "need" to develop it because of money.

    Imagine the following:
    Had AMD not pushed Mantle as far as they did giving developers exactly what they've been screaming for for years, would you have DX12 right now?
    Very likely you would, but you would not have control over the resources of the GPU as much as you would have now.

    In fact I am pretty sure that DX12 would've been far less control over the GPU's resources than it is now.
    So does this mean that dx12 will help level the playing field even more?

  9. #9
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Lightwysh View Post
    So does this mean that dx12 will help level the playing field even more?
    Potentially, not necessarily.

    This will depend fully upon the developers of said application/game but yes the potential is there.

    However like said before do not attempt to compare Intel's 8-core vs. AMD's 8-core in an ideal scenario with DX12 f.ex.
    If the graphics had no limits etc. and it was purely limited to CPU bottlenecking the Intel would still annihilate AMD's 8-core.

    This is easily shown for video rendering f.ex. that use the CPU.. since that has little to nothing to do with GPU. (in most programmes at least)

    AMD's 8-core should ideally be compared to Intel's quad cores.

    That is untill the Zen architecture CPUs from AMD emerge... IF they can stick to what they are gunning for and promised.

  10. #10
    Yeah, the reason it wasn't making an awful lot of sense in my head was that I assumed the CPUs were being stressed to some degree.

    Anyway, not getting into the processors themselves here, I'm still sticking to my guns on this front: I want more motherboards for AMD. Jesus, it's still the AM3+... Intel at least leveraged me to upgrade because I wanted the newer features, more than I cared about the raw performance upgrade for the processor.

  11. #11
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Quote Originally Posted by Lightwysh View Post
    Can we have a discussion on dx12 amd vs intel logically without a heavy dose of bias on this forum?
    Probably not.

    For the best result for DirectX12 atm is AMD GPU and Intel CPUs. Remember that any CPU benefit for AMD's CPU also will benefit Intel's too. So it's not like it's one vendor bias for this. NVidia's does not use Asynchronous compute in this game so this actually isn't even a true apples to apples comparison. An Oxide developer(of this game) noted it as "an unmitigated disaster".
    Benchmark from PCPer, not the greatest review site due to being basically NVidia PR but it's the only one I can remember.
    http://www.pcper.com/reviews/Graphic.../Results-Avera
    Last edited by Remilia; 2015-09-03 at 04:59 PM.

  12. #12
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    That seems awkwardly skewed, the AMD CPUs should be performing better than that with their potential horsepower.

    Considering there are 8 threads it should be pushing more as you can see in video rendering benchmarks.

    Would be a shame to see that much horsepower go to waste.

  13. #13
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Quote Originally Posted by Evildeffy View Post
    That seems awkwardly skewed, the AMD CPUs should be performing better than that with their potential horsepower.

    Considering there are 8 threads it should be pushing more as you can see in video rendering benchmarks.

    Would be a shame to see that much horsepower go to waste.
    Games work differently than video rendering. Not to also mention the FPUs doesn't really translate well into CPU heavy games. Higher IPC still trumps it despite it being core + SMT as opposed to an FPU.

  14. #14
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Remilia View Post
    Games work differently than video rendering. Not to also mention the FPUs doesn't really translate well into CPU heavy games. Higher IPC still trumps it despite it being core + SMT as opposed to an FPU.
    Granted.. however there should still be more horsepower available than what is shown there.

    Taking the i3-4330 for example is still 2 cores and hyperthreading which leverage IPC at exactly the same rate prior or post DX12.
    A dual core should not have more overall horsepower when the scope of use was the limited amount of processing power available due to cores being addressed.

    An Intel CPU would of course gain just as much as an AMD CPU in this regard but due to the amount of cores the potential power of the CPU should be higher than it is.

    The whole idea of DX12 would be to remove the limit on the CPU and pass it to the GPU (or as close to as possible) so I'm having a hard time believing that a dual core from Intel (as good as they are though) can trounce the FX-8370 regardless of architectural differences.

    If however the architecture is wholly unsuited for DX12 than that sucks for AMD's FX series and their APU series as well since they will have A LOT riding on this.

    The APUs are also 1 of the reasons I think something is off regarding that benchmark.

  15. #15
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    One thing about DX12/Vulcan/Mantle isn't to remove the CPU limit, it's to reduce the CPU overhead. That's not the same as shifting more burden on the GPU. Tasks assigned to the CPU are still assigned to the CPU. This allows for more things to be shoved onto the CPU. If a game requires more CPU power, it'll benefit from a faster IPC. You can't split one task into different cores, but you can assign different tasks to different cores much better now. That one task is still bound by the speed of the cores.
    The newer APIs have more than just CPU overhead reduction though. It's just one of the few things like asynchronous compute and multi-gpu rendering.

  16. #16
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Remilia View Post
    One thing about DX12/Vulcan/Mantle isn't to remove the CPU limit, it's to reduce the CPU overhead. That's not the same as shifting more burden on the GPU. Tasks assigned to the CPU are still assigned to the CPU. This allows for more things to be shoved onto the CPU. If a game requires more CPU power, it'll benefit from a faster IPC. You can't split one task into different cores, but you can assign different tasks to different cores much better now. That one task is still bound by the speed of the cores.
    The newer APIs have more than just CPU overhead reduction though. It's just one of the few things like asynchronous compute and multi-gpu rendering.
    Yeah and part of the DX12 / Mantle / Vulkan directive was to shift some calculations from the CPU to the GPU hence the compute pipelines in them.

    I may have phrased my sentence wrongly, I should phrased it as thus:

    "The whole idea of DX12 would be to shift the idea of the CPU being a bottleneck to the idea of data being processed and calculated on the GPU rather than CPU, such as physics for example."

    This is why larger impacts on performance is seen on f.ex. dual core CPUs than octa-core CPUs, this was the case for Mantle as well.

    But in the end when accounting for a state wherein a lot of commands are processed by the CPU unilaterally the 8-core in an ideal situation should beat the dual core don't you think? Especially since Ashes of the Singularity along with DX12 should allow for draw calls to be request from all CPU cores at the same time.

  17. #17
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Except it's not an 8 core, it's a 4 core + 4 FPU, which is weaker than a full core and specializes in floating point calculation. Ashes utilizes a lot of AI calculations though, something that can't be done on GPU compute. GPU compute can do somethings, things that don't require CPU interactions, aka post processing.

  18. #18
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Remilia View Post
    Except it's not an 8 core, it's a 4 core + 4 FPU, which is weaker than a full core and specializes in floating point calculation. Ashes utilizes a lot of AI calculations though, something that can't be done on GPU compute. GPU compute can do somethings, things that don't require CPU interactions, aka post processing.
    Well technically they are 8 real cores just how they are set up kinda wrecks performance when it's fully stressed due to the fact that 1 module of 2 cores share 1 FPU.
    So it's not really 4 cores and 4 FPUs, more like 2 cores share 1 FPU in 4 modules.
    A random source regarding core/FPU division

    I believe I read an article somewhere that they switched to automated design for the FX series CPUs where if it were manually designed it would be 20% smaller, more efficient and higher IPC values.. but for the life of me can't find it now.

    Regardless AI calculations are AFAIK separated to share equally per CPU core (I MIGHT have read that somewhere regarding AotS, not sure) so the spread should still be equal... who knows..

    Maybe AMD's FX chips are just unsuited as a whole for DX12 as well but honestly considering what they have riding on it, including their APUs... seems off.
    Perhaps someone should ask that in a tweet or something to the devs of AotS regarding FX line of CPUs, I'm curious to know.

    (I don't use any social media... I hate it, I cannot explain it other than me thinking it's a disease to humanity)

  19. #19
    Watch this for the subject of driver overhead.

  20. #20
    Quote Originally Posted by Evildeffy View Post
    Potentially, not necessarily.

    This will depend fully upon the developers of said application/game but yes the potential is there.

    However like said before do not attempt to compare Intel's 8-core vs. AMD's 8-core in an ideal scenario with DX12 f.ex.
    If the graphics had no limits etc. and it was purely limited to CPU bottlenecking the Intel would still annihilate AMD's 8-core.

    This is easily shown for video rendering f.ex. that use the CPU.. since that has little to nothing to do with GPU. (in most programmes at least)

    AMD's 8-core should ideally be compared to Intel's quad cores.

    That is untill the Zen architecture CPUs from AMD emerge... IF they can stick to what they are gunning for and promised.
    Seems like the most important comparison is price.
    While you live, shine / Have no grief at all / Life exists only for a short while / And time demands its toll.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •