[gamegpu.ru] APU gaming including Skylake GT2 and Broadwell Iris Pro 6200

[Arachnotronic]

What I find funny is that as good as Skylake is about advanced media codec support, the top mobile processors/GPUs are often ahead. SD820 for example supports HEVC Main 10 in hardware.

 

[ShintaiDK]

Its the cycle time for development. See Apple as the opposite, stuck at 264 with AppleTV for example.

 

[mikk]

Re-read post #84, dude. Scores posted -- Respective Haswell Pentium scores lower than the Kaveri in single threaded benchmark according to Passmark. Don't know how you could be any more dense. If Phynaz's earlier comment
was actually true, what I linked would be impossible.

Read carefully

He said Kaveri can't have the same single thread performance which you can't disconfirm with just one benchmark, even less when it's a synthetic benchmark.

You don't understand this it seems. With you logic I could claim that FX-8350 isn't slower per thread as a 6700k if I could find one single odd benchmark where this is the case. You have to do more than just posting a random benchmark score from a synthetic benchmark. As I said you didn't prove anyone wrong.

 

[jihe]

A totally wrong blanket statement.

Passmark Single Threaded Scores (respectively):
Intel Pentium 3220T / Haswell 1150 = 1434
https://cpubenchmark.net/cpu.php?cpu=Intel+Pentium+G3220T+@+2.60GHz&id=2189

Kaveri A10 7850K / Quad Core FM2+ = 1567
https://cpubenchmark.net/cpu.php?cpu=AMD+A10-7850K+APU&id=2133

You know Intel makes more than one Haswell Pentium and AMD makes more than one Kaveri, right? Clearly not.
There are quite a few Kaveri chips that outperform several Haswell Celeron and Pentium chips in single threaded performance.
Kaveri does needs a little extra clock speed to do it, but your statement is factually incorrect.

My overclocked 860K will thrash pretty much anything short of an i3 in single threaded performance. Obviously,
My 4790K kills it, the AMD is not even remotely close -- but considering my i7 was 3 times the cost, the little 860K
is a very solid chip.

3220T? Really? Why don't you just put up some haswell mobile scores? Or perhaps the 1.6g haswell xeons?

 

[AtenRa]

Two more games

Pity they havent benchmarked from 720p to 1080p as they did before.

 

[frozentundra123456]

Why would you test Just Cause 3 at 1080p? Nothing except iris pro can even do 30 fps at 720p. Just goes to show how weak igpus from either AMD or intel really are for desktop gaming.

 

[beginner99]

At 1080p on those two games, it seams that pushing image quality has a higher performance impact on Broadwell Iris Pro 6200 performance allowing Kaveri 65W TDP A8-7600 to close the gap.

That's because here you see how the AMD APUs lack CPU performance because they are already CPU limited at that quality setting compared to the broadwell Iris Pro.

Has nothing to do with quality just CPU vs GPU limit. That's why APUs don't lose much performance. On a positive note you could say for this selection of games they are better balanced but then no sane person will game on a broadwell Iris igpu. You buy it because of the edram and then use a dGPU.

 

[Gikaseixas]

AMD needs faster CPU cores to fully utilize the GPU, otherwise Just Cause 3 will be first of many more titles AMD fails flat on it's face.

Zen based APU's better deliver

 

[parvadomus]

AMD needs faster CPU cores to fully utilize the GPU, otherwise Just Cause 3 will be first of many more titles AMD fails flat on it's face.

Zen based APU's better deliver

I dont think so.

Maybe that game is just too intesive for the APU, in both cpu and gpu fronts, and it just starts to throttle. Maybe its just drivers.
The other AMD CPUs perform OK with that game.

 

[Gikaseixas]

I dont think so.

Maybe that game is just too intesive for the APU, in both cpu and gpu fronts, and it just starts to throttle. Maybe its just drivers.
The other AMD CPUs perform OK with that game.

Uh?

FX 9590 needs to operate at 1200 mhz higher than a 4770k to come close to it. Intel got some powerful cores period.
I have both arch's @ 4.6ghz, it's quite clear who wins. I tested many apps and games.

 

[parvadomus]

Uh?

FX 9590 needs to operate at 1200 mhz higher than a 4770k to come close to it. Intel got some powerful cores period.
I have both arch's @ 4.6ghz, it's quite clear who wins. I tested many apps and games.

Everyone knows that. But you missed the point obviously.
If AMD CPUs can run the game OK with a 980TI. I dont understand how they can bottleneck a low end 512sp running DDR3 memory. Its obvious that the problems with just cause 3 with the APU are beyond the IPC deficit.

 

[MiddleOfTheRoad]

3220T? Really? Why don't you just put up some haswell mobile scores? Or perhaps the 1.6g haswell xeons?

Talk about a necro thread. This was pretty much sorted out weeks ago. I'm not going to continue to beat this dead horse. He spoke in general terms and I pointed out there were clear exceptions to what he state....

There is really nowhere to go from there.

 

[AtenRa]

Why would you test Just Cause 3 at 1080p? Nothing except iris pro can even do 30 fps at 720p. Just goes to show how weak igpus from either AMD or intel really are for desktop gaming.

Or how badly optimized Just Cause is when other games with comparable or better graphics can produce more than 30fps at 1080p on APUs.

 

[frozentundra123456]

Yes, but the cpus in the gpu test are not FM2, plus they dont have to share TDP with the igpu. The fact that Skylake HD530 is as fast as the fastest AMD APU, with much better minimums, makes it seem very likely the AMD APUs are cpu limited.

 

[AtenRa]

That's because here you see how the AMD APUs lack CPU performance because they are already CPU limited at that quality setting compared to the broadwell Iris Pro.

Has nothing to do with quality just CPU vs GPU limit. That's why APUs don't lose much performance. On a positive note you could say for this selection of games they are better balanced but then no sane person will game on a broadwell Iris igpu. You buy it because of the edram and then use a dGPU.

CPU performance at 1080p ??? every APU and Intel iGPU is GPU limited at 1080p, CPU performance is more than enough for those iGPUs even at lower resolutions.

 

[AtenRa]

Yes, but the cpus in the gpu test are not FM2, plus they dont have to share TDP with the igpu. The fact that Skylake HD530 is as fast as the fastest AMD APU, with much better minimums, makes it seem very likely the AMD APUs are cpu limited.

or a driver/game optimization is needed.

 

[frozentundra123456]

Or how badly optimized Just Cause is when other games with comparable or better graphics can produce more than 30fps at 1080p on APUs.

The point is, this game is unplayable on the AMD APUs. Even if you can live with upper 20s as the average framerate, the minimums make it unplayable. Upping the resolution will not increase performance.

 

[AtenRa]

The point is, this game is unplayable on the AMD APUs. Even if you can live with upper 20s as the average framerate, the minimums make it unplayable. Upping the resolution will not increase performance.

Nobody said the game is playable at those settings, i would really like to see how all the APUs,AMD and Intel, would perform at 900p and 1080p. That is what im saying.

 

[Gikaseixas]

Everyone knows that. But you missed the point obviously.
If AMD CPUs can run the game OK with a 980TI. I dont understand how they can bottleneck a low end 512sp running DDR3 memory. Its obvious that the problems with just cause 3 with the APU are beyond the IPC deficit.

Those APU's have two modules, not four like the FX's and from the benchmarks this is a well threaded game, the more cores your cpu has the better, plus the faster the core even better

The only one who's clearly missing points here it's you

Two AMD modules and lack of bandwidth (DDR3 dependent) is not enough.

 

[Gikaseixas]

Nobody said the game is playable at those settings, i would really like to see how all the APUs,AMD and Intel, would perform at 900p and 1080p. That is what im saying.

I think he knows very well what you meant. I would like to see those tests too, just for the sake of knowledge. Why defend limited tests? does not compute

 

[Gikaseixas]

Yes, but the cpus in the gpu test are not FM2, plus they dont have to share TDP with the igpu

Don't always agree with you but you're correct on this point sir

 

[AtenRa]

Those APU's have two modules and from the benchmarks this is a well threaded game, the more cores your cpu has the better, plus the faster the core even better

The only one who's clearly missing points here it's you

Two AMD modules and lack of bandwidth (DDR3 dependent) is not enough.

Im not convinced the performance problem is related to the weak CPU. Have a look at the Dragon Dogma : Dark Arisen cpu scaling, it can use more than 4 cores/threads and yet the APUs are doing more than fine.

 

[Gikaseixas]

Im not convinced the performance problem is related to the weak CPU. Have a look at the Dragon Dogma : Dark Arisen cpu scaling, it can use more than 4 cores/threads and yet the APUs are doing more than fine.

Don't take me wrong but you need to be selective in order to enjoy gaming on APU's, this just proves that once again.

Perhaps Just Cause exposes APU's limits more than other titles so let's agree to disagree

 

[AtenRa]

Don't take me wrong but you need to be selective in order to enjoy gaming on APU's, this just proves that once again.

Perhaps Just Cause exposes APU's limits more than other titles so let's agree to disagree

Definitely, we cannot play every game with an APU that is a fact. And Just Cause 3 may be one of them.
But i would really like to see a review again after a few drivers/game patches have been released.

 

[Yuriman]

Definitely, we cannot play every game with an APU that is a fact. And Just Cause 3 may be one of them.
But i would really like to see a review again after a few drivers/game patches have been released.

Which is a shame. I think APUs will "be there" when they can play all games, even if you need to use minimum settings in some. Having some games be unplayable at *any* settings is an issue.