AMD responds to 1080p gaming tests on Ryzen. Supports ECC RAM. Win 10 SMT bug

[zMeul]

5 minutes ago, XenosTech said:

What you say makes as much sense as partick calling the zen chip a xeon

mate, I asked you to point me specifics of what I said that doesn't sit well with you

maybe we'll sort it out

 

but all I see from you is randomness about other forum member I have nothing in common with

so, care to get back on topic? or shall I ignore you from now on?

[Guest Kloaked]

2 minutes ago, Zi6iX said:

I do believe that the low performance in gaming is due to some intel optimization. I mean look at cinebench, CPU-Z, 3d mark, and all of those neutral test they're pretty much on par with intel. We also know that the ipc about 6% lower than sky/kabylake.

People keep claiming software bias but I don't buy it. AMD has continually used this excuse and it's getting old.

 

Which tests are "neutral"?

[Memories4K]

Quote

A from AMD_LisaSu : Ryzen is doing really well in 1440p and 4K gaming when the applications are more graphics bound.

[Curious Pineapple]

2 minutes ago, Zi6iX said:

I do believe that the low performance in gaming is due to some intel optimization. I mean look at cinebench, CPU-Z, 3d mark, and all of those neutral test they're pretty much on par with intel. We also know that the ipc about 6% lower than sky/kabylake.

But that's irrelevant here. It's easier to ignore the facts and keep spouting how much better Intel is at everything. Everyone here seems to be experts at CPU design and can spin something better out of their arse.

[XenosTech]

8 minutes ago, zMeul said:

mate, I asked you to point me specifics of what I said that doesn't sit well with you

maybe we'll sort it out

 

but all I see from you is randomness about other forum member I have nothing in common with

so, care to get back on topic? or shall I ignore you from now on?

 

1 hour ago, zMeul said:

AMD blaming Intel for their poor performance in gaming .. yes, yes, what's new?

 

but wait! weren't those same compilers used for the software where Zen showed to be better than Broadwell - yeah, that's what I thought

 

1 hour ago, zMeul said:

the same compilers were used in the cases where Zen beats Broadwell, so ... hmm

 

3 hours ago, zMeul said:

1800X is more expensive that 7700K, 1700X is still more expensive than 7700K

1700 is cheaper, 10$, than 7700K but also slower than 7700K

what do you chose? I know .. the A10-7890K because it's not in the margin of error 

This one takes the cake though

3 hours ago, zMeul said:

except the 7700k beats it:

 

having more cores does not equate to having more performance

I really can't do this all day, I have work to do. Networks to fix, server deployments to do

[AnonymousGuy]

Let the damage control begin!  It's like AMD has a flow chart of excuses when their engineering can't cash the checks their marketing writes.  Been like this forever.

[VagabondWraith]

2 minutes ago, Memories4K said:

I mean shit, a Core 2 will do good at 4K.  

[zMeul]

7 minutes ago, XenosTech said:

I really can't do this all day, I have work to do. Networks to fix, server deployments to do

1. the compiler developers use for <insert random software name> are the same with what game developers use - to claim that game developers use "special" compilers provided by Intel to skew the results in their favor is utter bullshit; this is especially damning on games that AMD worked closely with the game developers and showed poor performance with Zen
2. ^ takes care of this too
3. it was soft of a joke, don't recall exactly the context
4. yes, having more cores does not translate in better performance - just look at the 6900K vs 7700K in gaming, for example; WatchDogs2 is a game that uses all the cores of the 1800X and yet it still gets stomped by the 7700K

[XenosTech]

1 minute ago, zMeul said:

1. the compiler developers use for <insert random software name> are the same with what game developers use - to claim that game developers use "special" compilers provided by Intel to skew the results in their favor is utter bullshit; this is especially damning on games that AMD worked closely with the game developers and showed poor performance with Zen
2. ^ takes care of this too
3. it was soft of a joke, don't recall exactly the context
4. yes, having more cores does not translate in better performance - just look at the 6900K vs 7700K in gaming, for example; WatchDogs2 is a game that uses all the cores of the 1800X and yet it still gets stomped by the 7700K

The mere fact that all this time you were jumping up and down saying the 1800x was bad because it loses to the 7700k but now you finally acknowledge that the 6900k is getting creamed by the 7700k... Interesting

[zMeul]

Just now, XenosTech said:

6900k is getting creamed by the 7700k... Interesting

was this not the case!? did you saw me say otherwise?!

[Hunter259]

24 minutes ago, M.Yurizaki said:

Not really number crunching. It's more like it shifted the burden of providing work for the GPU from the API to the application. The API opens more channels and defines new data sets so the driver can compile them into GPU tasks faster.

 

However that actually makes it potentially more sensitive to more cores. But at the moment, most games only implement three workload queues, which three threads can handle easily.

 

Yes. x86 only defines the operation, the input, and the expected output. It does not define how it should be implemented. Some operations on Ryzen can be faster or slower than the same operation on Skylake.

Technically all processors do is number crunching

[MakeAMDGreatAgain]

20 minutes ago, Memories4K said:

lol did she really say that?

[Curious Pineapple]

1 minute ago, zMeul said:

1. the compiler developers use for <insert random software name> are the same with what game developers use

The same code compiled for different ARM cores is optimized for the target processor. It will still run on multiple variants, but with lower performance. x86 is the same, code can be targetted towards either Intel or AMD, or you can have a more generic output code that takes advantage of no optimisations, but is guaranteed to work just as badly on either.

 

There is likely no compiler in widespread use currently that can compile code which takes advantage of Zen. In fact, any AMD optimisation is likely to be aimed towards Bulldozer, and may assume only 4 FPU's are available when I believe that Ryzen 8c/16t has 16 FPU's.

[Memories4K]

2 minutes ago, MakeAMDGreatAgain said:

lol did she really say that?

Check this out lad

[MakeAMDGreatAgain]

3 minutes ago, Memories4K said:

Check this out lad

lol. I'm going to add that to my signature as a quote. ty

[zMeul]

6 minutes ago, Curious Pineapple said:

There is likely no compiler in widespread use currently that can compile code which takes advantage of Zen. In fact, any AMD optimisation is likely to be aimed towards Bulldozer, and may assume only 4 FPU's are available when I believe that Ryzen 8c/16t has 16 FPU's.

Microsoft's compilers are the ones used and MS' compilers are agnostic

MS' compilers for Windows are centered around the x86 arch - why are you even bringing up ARM!?

x86 was developed by Intel .. tough luck

 

if AMD makes changes in the logic of the x86 arch, it's on them

what MS can do is to issue microcode updates for the kernel to work around the issues - Zen has been 5y years in development and you're telling me that only today they found out they're having issues?!

[FratStar]

52 minutes ago, MyName13 said:

Is software optimization for an x86 CPU architecture a thing?Isn't software supposed to be optimized to use additional cores and threads instead?

Yes I've done a small bit of game engine modification myself. Jumping into the source code of id tech engines, source, etc. The code detects the cpu brand and run specific operations and code to do the same thing and it may be faster for. Intel or AMD, but they yield the same/similar results.

[xAcid9]

1 minute ago, goodtofufriday said:

I Feel like this needs to be here too.

 

-snip-

Can you kindly Ctrl+A > Ctrl+X > Ctrl+Shift+V or Ctrl+A and then click Remove Formating everytime you want to retoast that?    

 

 

[goodtofufriday]

Just now, xAcid9 said:

Can you kindly Ctrl+A > Ctrl+X > Ctrl+Shift+V or Ctrl+A and then click Remove Formating everytime you want to retoast that?    

 

Lol Is it coming out white on the dark theme?

[DocSwag]

2 hours ago, Curufinwe_wins said:

Well that and AMDs alu/fp units are incredibly weak compared to Intel's right now...

 

For most consumer workloads it doesn't matter (more so because of poor software optimization), but it is quite reasonable to see some bottlenecks in things that should rely heavily on that (like draw call limited gaming).

 

Also yea... Zen is NOT a HPC cpu. Unfortunately...

 

I mean it wasn't unexpected... but it is a bit disappointing to see just how much weaker it is in those workflows...

 

Still great launch by AMD for consumers.

I wouldn't be surprised if in their next iteration of Zen they put in a larger FPU to make it more suitable for HPC. They've said they know of a lot of small additions that they want to make with the next iteration of Zen so I wouldn't be surprised if the next iteration has a 2x256 bit FMAC.

[Curious Pineapple]

9 minutes ago, zMeul said:

Microsoft's compilers are the ones used and MS' compilers are agnostic

MS' compilers for Windows are centered around the x86 arch - why are you even bringing up ARM!?

x86 was developed by Intel .. tough luck

 

if AMD makes changes in the logic of the x86 arch, it's on them

what MS can do is to issue microcode updates for the kernel to work around the issues - Zen has been 5y years in development and you're telling me that only today they found out they're having issues?!

I'm bringing up ARM because it's the same thing. There is a whole lineup of ARM based processors out there, each different, and code can be optimised for each of them. x86/x86-64 is the same. 2 different companies with products based on the same base instruction set/arch, each designed differently, and will react differently to the same code.

 

Zen has been in development for 5 years yes, but how long have they had actual working chips for? How many different wafers have they been through? It takes/can take months to actually go from raw silicon wafers to a packaged CPU. They have to at some point just start making them, bugs or not. If they kept making changes and waiting until it was the perfect CPU and was as good as it can be, it will never be released. When laying out 2 billion transistors in silicon, and any bugs are either minor or fixable without a hardware change, it's good to go. CPU design isn't programming, it's basic electronics on a nanometer scale.

[zMeul]

1 minute ago, Curious Pineapple said:

Zen has been in development for 5 years yes, but how long have they had actual working chips for? How many different wafers have they been through? It takes/can take months to actually go from raw silicon wafers to a packaged CPU. They have to at some point just start making them, bugs or not. If they kept making changes and waiting until it was the perfect CPU and was as good as it can be, it will never be released. When laying out 2 billion transistors in silicon, and any bugs are either minor or fixable without a hardware change, it's good to go. CPU design isn't programming, it's basic electronics on a nanometer scale.

as I said previously, they didn't just stumbled on the finished product yesterday

no one forced them to release the CPUs without verifying everything and pushing the updates to the right developers

[Curious Pineapple]

1 minute ago, zMeul said:

as I said previously, they didn't just stumbled on the finished product yesterday

no one forced them to release the CPUs without verifying everything and pushing the updates to the right developers

OK, let's put this another way. The god damn thing was released today, reviewers have had them for longer. Let's assume that AMD have given Microsoft all they need to produce code that performs better on Ryzen. VS 2017 is 5 days away, are they really going to update, rebuild and push out a new version of an older toolchain a week before they release the replacement version?

[zMeul]

1 minute ago, Curious Pineapple said:

OK, let's put this another way. The god damn thing was released today, reviewers have had them for longer. Let's assume that AMD have given Microsoft all they need to produce code that performs better on Ryzen. VS 2017 is 5 days away, are they really going to update, rebuild and push out a new version of an older toolchain a week before they release the replacement version?

and you are assuming people will just top using VS2015 because?!

 

mainstream support ends 10/13/2020; extended support ends 10/14/2025  https://support.microsoft.com/en-us/lifecycle/search/19591

there goes your excuse out of the window 

[Prysin]

Lol. I told you Ryzen would be a disappointment. I told you back in November.