AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

[frozentundra123456]

Well I'm excited to see what ZEN has to offer. As long as it is faster than my 2600K and doesn't need a nuclear power plant to run, I'm in.

Why? There are plenty of cpus that will do that now. 6700k for faster single thread or HEDT for better multithreaded.

 

[sirmo]

Strange issue!

It's pretty well known with 1st gen Haswells http://www.overclock.net/t/1490212/weird-issue-with-4770k-delid-massive-overheating

 

[krumme]

Why? There are plenty of cpus that will do that now. 6700k for faster single thread or HEDT for better multithreaded.

The reason the rest of us just dont buy a 8 core bwe with an expensive mb. A total of what 1400 usd, where a zen 8c system might be half that. Only cost matters here.

 

[cytg111]

Why? There are plenty of cpus that will do that now. 6700k for faster single thread or HEDT for better multithreaded.

Normally I'd agree but that BF1 CPU usage chart showing a 3GHz 5960 owning everything has me pausing for a moment.
Personally I didnt see us moving past quads for another decade .. right now I am cautiously optimistic

 

[dfk7677]

BF1 should be used as a multithreading benchmark

 

[Atari2600]

Personally I didnt see us moving past quads for another decade .. right now I am cautiously optimistic

For gaming, yes. Probably maybe.

But for some folks, the whole world doesn't revolve around FPS in battlefield!

 

[frozentundra123456]

The reason the rest of us just dont buy a 8 core bwe with an expensive mb. A total of what 1400 usd, where a zen 8c system might be half that. Only cost matters here.

Yea, and if Zen only matches SB ipc, as the poster I replied to said was sufficient, and probably at lower clocks, it will only match hex core HEDT which is available for 400 dollars. That was the point.

 

[TheELF]

Are you guys talking about the gamegpu BF1 results?
Cause those aren't really that impressive for the 8/16 intel chip,granted its gpu bottlenecked even with the best gpu available at 1080p but the 4c i5 at 3,3 gets ~90fps min if your argument is that the 5960 gets used to a high degree then the i5 is being used close to 100% so any oc will be translated to FPS almost 100% and you can sky oc even the locked i5 to at least around 4,5 that's a ~30% increase, that would give ~30% more FPS bringing the minimums up to around 120 so nothing really gained from 4 cores more and 8 ht more.
Unless you lower detail/quality or resolution but then you'd need new benchmarks.

 

[cytg111]

Sorry, brain went the other way, I meant watch dogs 2, not bf1.

 

[TheELF]

Sorry, brain went the other way, I meant watch dogs 2, not bf1.

If you look at watch dogs 1 benchmark,and read between the lines,you will see that exactly the same was happening there, there just weren't any GPU around that were fast enough to show the same difference as today.

2600k and 4770k practically on the same level and the 3970 ~10fps more with every core and thread working,just no GPU that can show the difference even in sli.
This is more of an crappy engine/utilization then it is good multithreading.

 

[cytg111]

For gaming, yes. Probably maybe.

But for some folks, the whole world doesn't revolve around FPS in battlefield!

And for those particular folks a dual core-m should be more than enough .

 

[deasd]

Yea, and if Zen only matches SB ipc, as the poster I replied to said was sufficient, and probably at lower clocks, it will only match hex core HEDT which is available for 400 dollars. That was the point.

If 8C16T match 6c12T, the 8C16T might cost less.

 

[bjt2]

Yea, and if Zen only matches SB ipc, as the poster I replied to said was sufficient, and probably at lower clocks, it will only match hex core HEDT which is available for 400 dollars. That was the point.

Why lower clocks? XV on the 28nm already has 4.3GHz turbo... Why Zen should be lower? It has also at least the same pipeline length (and thus the same FO4)...

 

[TheELF]

Why lower clocks? XV on the 28nm already has 4.3GHz turbo... Why Zen should be lower? It has also at least the same pipeline length (and thus the same FO4)...

XV is 4 cores max and not 8/16 and from what we have seen so far amd is much more concerned about power this time around.

 

[Arachnotronic]

edit: nevermind.

 

[AtenRa]

Lets wait and see the performance vs price and then we can talk about.

 

[Atari2600]

And for those particular folks a dual core-m should be more than enough .

Erm, as Roy Walker would say, its good, but its not right.

I was thinking more of prosumers on a budget. Or even folks at home looking to do some media encoding work. Not spend multiple thousands on a machine, but something with good throughput on the rare occasions its asked to do some heavy lifting.

 

[TheELF]

Or even folks at home looking to do some media encoding work.

Ever since haswell, qsv allows you to convert 1080p@60fps x.264 videos at +200fps even on the slowest celeron, kaby is supposed to do the same for x.265.
General public doesn't care about pixel perfect quality at the smallest possible file size, they just use whatever preset is default and only ever complain if they get really huge files.
Now that is Accelerated Processing at it's finest.

 

[Atari2600]

Ever since haswell, qsv allows you to convert 1080p@60fps x.264 videos at +200fps even on the slowest celeron, kaby is supposed to do the same for x.265.
General public doesn't care about pixel perfect quality at the smallest possible file size, they just use whatever preset is default and only ever complain if they get really huge files.
Now that is Accelerated Processing at it's finest.

I guess that's a fair point for that specific example I've mentioned.

Engineering codes etc are too niche to be considered relevant. The likes of blender etc may be more relevant, but, same as engineering, the market size balancing between performance:cost is very small.

 

[bjt2]

XV is 4 cores max and not 8/16 and from what we have seen so far amd is much more concerned about power this time around.

They said that Zen has the same energy consumption that excavator (obviously at same clock) core to core, so at least the 4c/8t should arrive at 4.1-4.3GHz in 95W, but that was an APU (so also GPU is in the TDP)... For the 8 core we should stick at the old 32nm SOI and the 8c FX 8370E: 3.3-4.3GHz at 95W or the 8370 at 4-4.3GHz in 125W... The 8c/16t probabily will go over BW-E clocks, especially if a 140W SKU is ever produced...

 

[Atari2600]

They said that Zen has the same energy consumption that excavator

They did?

Any reference for that, including this bit:

(obviously at same clock) core to core,

please?

 

[sirmo]

Ever since haswell, qsv allows you to convert 1080p@60fps x.264 videos at +200fps even on the slowest celeron, kaby is supposed to do the same for x.265.
General public doesn't care about pixel perfect quality at the smallest possible file size, they just use whatever preset is default and only ever complain if they get really huge files.
Now that is Accelerated Processing at it's finest.

AMD is adding VCE support to Handbrake as well. https://subtletv.com/baab5YY/Handbrake_now_supports_AMD_VCE_Acceleration

However I doubt summit ridge will have the VCE.. APUs and AMD GPUs however do. However if you're talking about professional video editors, more cores is a win win, because video editing is not only about transcoding.

 

[sirmo]

Excavator power consumption is actually quite good. Carrizo has some really nifty power management features (AVFS). Perf/watt isn't great because it's still based on Bulldozer, but if Zen can have same power consumption clock for clock as Excavator while boosting the IPC by 40%+ than it's big win. And it might give intel run for its money in perf/watt.

 

[KTE]

They did?

Any reference for that, including this bit:



please?

He has some pretty darn crazy, non scientific, theories comparing an ARM research test FPU vs XV, applied to Zen in a sounds-good-to-me biblical zeal fashion.

In short, nothing of Zens power@clocks scaling is known yet.

Nor how it compares to XV.

Nor how the process stacks up vs Intels.

Sent from HTC 10
(Opinions are own)

 

[mikk]

AMD is adding VCE support to Handbrake as well. https://subtletv.com/baab5YY/Handbrake_now_supports_AMD_VCE_Acceleration

This is just marketing from AMD, look how old this video is. Another case of "AMD likes to promote things in slides that they don't support in real wold"....only that it's a video

And VCE is poor compared to Quicksync and NVENC.