Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[DrMrLordX]

I am definitely in favor of AMD opening up more memory multipliers. bclk overclockers are already hitting memory speeds of DDR4-3800 with DIMMs that will reach those heights.

 

[imported_jjj]

Really nobody reported on this yet?
CPC https://twitter.com/CPCHardware/status/844618089618722816
"2nd Gen B1 ES for Ryzen 16C/32T on X399 platform incoming. Speed (and TDP) bumped to 3.1/3.6 GHz @ 180W on these samples. Still not QS."

 

[thepaleobiker]

Really nobody reported on this yet?
CPC https://twitter.com/CPCHardware/status/844618089618722816
"2nd Gen B1 ES for Ryzen 16C/32T on X399 platform incoming. Speed (and TDP) bumped to 3.1/3.6 GHz @ 180W on these samples. Still not QS."

Saw this last night on twitterworld. I couldnt follow CPC's other French-language tweets though

 

[imported_jjj]

Saw this last night on twitterworld. I couldnt follow CPC's other French-language tweets though

My French s very rusty but not that rusty.

"On peut déjà s'attendre à un TDP réel > 200W. Pour un nouveau step., c'est incertain. Sur Ryzen, le switch A0->B1 a eu lieu entre ES et QS."
We can already expect a real TDP > 200W. For a new stepping, it's uncertain. With Ryzen, the transition from A0 to B1 came between ES and QS.

"@Xstaze Qualification Sample. Les échantillons qui ont les même spécifications que les modèles commerciaux."
Qualification Sample. The samples that have the same specs as commercial products.

 

[unseenmorbidity]

Someone on reddit got 2x16gb working at 3200 mhz

Yesterday after work I got my system up and running on a test bench and went to work on finding a stable memory clock before touching anything else.

After several hours of hair-pulling just trying to get past 2667, I finally got 3200 to boot, albeit with cas 18 when my kit was rated for 14. Turns out that the other timings besides cas could be way tighter! I could boot into windows as tight as 18-11-11-21, but as expected that threw errors 2 seconds into memtest86. I did some loosening to see what passed 5 mins of memtest86 and then started an overnight run at 18-14-14-30 when I went to sleep. Woke up to no errors

https://valid.x86.fr/w192bb

https://i.imgur.com/TqR6Wvf.jpg

F4-3200C14D-32GTZKW on Gigabyte K7, 3/14 beta bios

SOC 1.2v , DRAM 1.4v , DDRVPP and termination manually set to their default voltages

edit: rearranged some words for clarity

 

[piesquared]

Has any website written an article about how Ryzen is fluid and smoother in games yet?

 

[ManyThreads]

I did some tests yesterday, timing different Photoshop processes specific to my usage, comparing a Ryzen 1800X @ 4.0 Ghz and my i7 3770K @ 4.4 Ghz. Both systems had 16GB RAM, lots of GPU power, & SSD's. Basically HDR creation and saving/exporting is what I do. Highly multi-threaded processes (exporting/saving & the merging/alignment stages of HDR creation) were much faster on the Ryzen, no surprise there. What did surprise me was that some actions were significantly faster on my 3770K. I thought that Ryzen's IPC increase over Ivy Bridge would have been enough to negate, or at least match, the 400Mhz difference in clock speed. Does that make sense? Or does even just a few hundred Mhz on old architecture trump 2-3 generations of IPC gains? I expected them to at the very least match.

Long story short, I was left wondering if I shouldn't wait for 6-core Coffee Lake assuming I could get 4.5-4.8Ghz or so out of it (the 6850K easily clocks to 4.4, so I would expect Coffee Lake 6c to be better yet). Then I could (potentially) have more than 4 cores at a significantly higher clock. Thoughts? I guess I am just trying to find some sort of crossover point where core count AND clock speed are the highest possible relative to other options. Maybe Ryzen's 6-core options will clock to 4.5 as well? Just thinking out loud at this point.

 

[imported_jjj]

Has any website written an article about how Ryzen is fluid and smoother in games yet?

That might be true for more cores so not just Ryzen in games that scale to more cores. But hard to say if it makes any difference at normal res not low res and high FPS.

The 7700k sprinnts in some areas due to high ST clocks where the load is not high but it's suffocated in other areas where the load is too much for 4 cores. End result being a larger variance beteewn scenes. This seems to be the case with BF1 DX11 and Crysis 3 but at low res so it only matters for folks that are into high FPS.

I did some tests yesterday, timing different Photoshop processes specific to my usage, comparing a Ryzen 1800X @ 4.0 Ghz and my i7 3770K @ 4.4 Ghz. Both systems had 16GB RAM, lots of GPU power, & SSD's. Basically HDR creation and saving/exporting is what I do. Highly multi-threaded processes (exporting/saving & the merging/alignment stages of HDR creation) were much faster on the Ryzen, no surprise there. What did surprise me was that some actions were significantly faster on my 3770K. I thought that Ryzen's IPC increase over Ivy Bridge would have been enough to negate, or at least match, the 400Mhz difference in clock speed. Does that make sense? Or does even just a few hundred Mhz on old architecture trump 2-3 generations of IPC gains? I expected them to at the very least match.

Long story short, I was left wondering if I shouldn't wait for 6-core Coffee Lake assuming I could get 4.5-4.8Ghz or so out of it (the 6850K easily clocks to 4.4, so I would expect Coffee Lake 6c to be better yet). Then I could (potentially) have more than 4 cores at a significantly higher clock. Thoughts? I guess I am just trying to find some sort of crossover point where core count AND clock speed are the highest possible relative to other options. Maybe Ryzen's 6-core options will clock to 4.5 as well? Just thinking out loud at this point.

Some tasks will care a lot about memory BW and latency so what memory settings?
Ryzen 6 cores is not gonna clock higher but ,in theory, you do have a CPU upgrade path and you could upgrade the CPU in the future without a mobo upgrade.
Not gonna help with ST perf but there is also the soon to arrive 16 cores AMD. One can assume 700-1000$ for the CPU, costly mobos but hopefully unlocked.
Next year maybe AMD catches up in ST but this year,if you care more about ST, Intel is ahead.

 

[beginner99]

Has any website written an article about how Ryzen is fluid and smoother in games yet?

Has there been any proof of this besides the BF1 case a certain users post all over the place?

It's like HiFi sector. Every measurement tells you there is no difference between the $10 and $1000 cable yet the sound has more vibrance and clarity. I mean you have to justify the purchase somehow.

Why could Ryzen be more fluid? Because there is a CPU bottleneck and if you constantly run at say 60 fps is feels smoother than say a 7700k that runs at 120 fps, then gets to a GPU heavy area and drops to 80 fps and then back to 120 fps. How could you fix this? Get a FreeSync or Gsync monitor. The issue with stutter and smoothness is that above a certain threshold what we perceive as unsmooth are changes in framerate and not a generally low framerate.

 

[imported_jjj]

Has there been any proof of this besides the BF1 case a certain users post all over the place?

It's like HiFi sector. Every measurement tells you there is no difference between the $10 and $1000 cable yet the sound has more vibrance and clarity. I mean you have to justify the purchase somehow.

Why could Ryzen be more fluid? Because there is a CPU bottleneck and if you constantly run at say 60 fps is feels smoother than say a 7700k that runs at 120 fps, then gets to a GPU heavy area and drops to 80 fps and then back to 120 fps. How could you fix this? Get a FreeSync or Gsync monitor. The issue with stutter and smoothness is that above a certain threshold what we perceive as unsmooth are changes in framerate and not a generally low framerate.

That's total and utter BS and on purpose as facts have no relevance to you and it's all about defending Intel. If only i wouldn't have just explained one post above yours why there is less variance in games that scale to more cores.
Let's try with pictures maybe that's easier for you.
Crysis 3 Ryzen (red) vs 7700k, the average FPS is similar but the 7700k has a wider variance between scenes.
Just imagine that the Ryzen is an 8 cores Intel so it doesn't tickle you the wrong way.

 

[ManyThreads]

That might be true for more cores so not just Ryzen in games that scale to more cores. But hard to say if it makes any difference at normal res not low res and high FPS.

Some tasks will care a lot about memory BW and latency so what memory settings?
Ryzen 6 cores is not gonna clock higher but ,in theory, you do have a CPU upgrade path and you could upgrade the CPU in the future without a mobo upgrade.
Not gonna help with ST perf but there is also the soon to arrive 16 cores AMD. One can assume 700-1000$ for the CPU, costly mobos but hopefully unlocked.
Next year maybe AMD catches up in ST but this year,if you care more about ST, Intel is ahead.

The 3770K Rig had 16GB (2 sticks) of DDR3 2300 RAM CAS 16 or 18 timings (I forget exactly, but I bought it in 2012 so it's old)

The 1800X Rig had 16GB (2 sticks) of DDR4 3000 RAM and CAS 14 1T

The thing that really surprised me was 400 Mhz on a 6 year old CPU was significantly faster than any IPC gains made over that same time period, based on my tests, for processes that weren't as multi threaded. My test wasn't perfect but it was a very fair and controlled "real world" test, using the same images and everything, doing exactly what I use it for. Anything heavily multi-threaded though and the Ryzen destroyed it, but that was in line with expectations.

I REALLY want to buy Ryzen haha but this test kind of took the wind out of my sails given that it is basically what I would be buying a new computer to help speed up. Some processes are faster but some are slower. Overall it's probably a bit faster or equal once you average both processes (HDR Merge & HDR Create). Exporting is of course massively faster, as expected.

This is my graph (A NEF is a Nikon RAW image file, in this case 36 megapixels each):

 

[Ajay]

Ryzen 6 cores is not gonna clock higher but ,in theory, you do have a CPU upgrade path and you could upgrade the CPU in the future without a mobo upgrade.
Not gonna help with ST perf but there is also the soon to arrive 16 cores AMD. One can assume 700-1000$ for the CPU, costly mobos but hopefully unlocked.

Just so ppl don't misread this, the 16 core AMD HEDT CPU uses a different socket than AM4 - so it's not an upgrade path (I know you mentioned it, just wanted to make it clear). I wonder if AMD will be offering 12 & 16 core variants. I'm sure there will be more than one clock rate offered. Interesting stuff, for sure.

 

[clocks11]

Has it been explained why there is a version of the 8core Ryzen in 65watts, but not the 6 core cpu?

 

[unseenmorbidity]

Taich review tweaktown Best performance award

It's a real review too, and not a worthless infomercial. I am impressed.

 

[AtenRa]

Has it been explained why there is a version of the 8core Ryzen in 65watts, but not the 6 core cpu?

R5 1600 is 65W TDP

 

[clocks11]

R5 1600 is 65W TDP

Didn't see that listed in the article below:
http://wccftech.com/retailers-selling-amd-ryzen-5-cpus-3-weeks-official-launch/

 

[Malogeek]

nevermind...

 

[imported_jjj]

The 3770K Rig had 16GB (2 sticks) of DDR3 2300 RAM CAS 16 or 18 timings (I forget exactly, but I bought it in 2012 so it's old)

The 1800X Rig had 16GB (2 sticks) of DDR4 3000 RAM and CAS 14 1T

The thing that really surprised me was 400 Mhz on a 6 year old CPU was significantly faster than any IPC gains made over that same time period, based on my tests, for processes that weren't as multi threaded. My test wasn't perfect but it was a very fair and controlled "real world" test, using the same images and everything, doing exactly what I use it for. Anything heavily multi-threaded though and the Ryzen destroyed it, but that was in line with expectations.

I REALLY want to buy Ryzen haha but this test kind of took the wind out of my sails given that it is basically what I would be buying a new computer to help speed up. Some processes are faster but some are slower. Overall it's probably a bit faster or equal once you average both processes (HDR Merge & HDR Create). Exporting is of course massively faster, as expected.

This is my graph (A NEF is a Nikon RAW image file, in this case 36 megapixels each):

No idea what's going on ,maybe someone that knows Photoshop better can figure out why HDR is so slow.The difference seems too high to be IPC related.

 

[IEC]

Has there been any proof of this besides the BF1 case a certain users post all over the place?

It's like HiFi sector. Every measurement tells you there is no difference between the $10 and $1000 cable yet the sound has more vibrance and clarity. I mean you have to justify the purchase somehow.

Why could Ryzen be more fluid? Because there is a CPU bottleneck and if you constantly run at say 60 fps is feels smoother than say a 7700k that runs at 120 fps, then gets to a GPU heavy area and drops to 80 fps and then back to 120 fps. How could you fix this? Get a FreeSync or Gsync monitor. The issue with stutter and smoothness is that above a certain threshold what we perceive as unsmooth are changes in framerate and not a generally low framerate.

1. It's not just Ryzen. You will also experience less variance and smoother gameplay with Intel's HEDT offerings over quad cores. We're already at the point where many games scale beyond quad cores. Of particular note is that Kaby Lake has better IPC than the Broadwell-E based HEDT processors but is losing its crown as the "gaming" processor already in a battery of 16 tests Computerbase did earlier this year:

Source: https://www.computerbase.de/2017-02/cpu-skalierung-kerne-spiele-test/#abschnitt_performancerating_in_full_hd

2. I worked hard enough in years previous that I don't have to justify my computing purchases. Computers are my most affordable hobby. If Ryzen was the same performance or lesser performance versus my Skylake 6700K rig I would have returned the 1800X for a refund when my Crosshair VI Hero motherboard bricked. Instead, I bought two more motherboards and a Ryzen 1700. I am that impressed by AMD's latest offering (even given the platform teething issues). These are the first AMD CPUs I've purchased in over a decade. Maybe you don't ever encounter stutters in your usage due to maxing out all your threads, but if you remember what it was like to hit a wall back in the single core and dual core days, it's pretty obvious when you do. It was to me. Having task manager open and seeing all 8 threads pegged and getting stutters because of it = my excuse to upgrade to something better than 4C/8T.

P.S. I buy $2 audio cables and use a decade+-old pair of Sony MDR-V6 headphones with Beyerdynamic pads to replace the OEMs that flaked away many years ago.

 

[imported_jjj]

Didn't see that listed in the article below:
http://wccftech.com/retailers-selling-amd-ryzen-5-cpus-3-weeks-official-launch/

That's the most unreliable source possible. Better sites and pre-order listings have the 1600 at 65W.

 

[Ryzenenthusiast]

Didn't see that listed in the article below:
http://wccftech.com/retailers-selling-amd-ryzen-5-cpus-3-weeks-official-launch/

LOL... Sourcing Wccftech...

 

[clocks11]

That's the most unreliable source possible. Better sites and pre-order listings have the 1600 at 65W.

Cool. Thanks for the non-snarky response.

 

[guachi]

Depending on how you weight the results of the graph the Ryzen chip can be calculated to be much faster.

Just adding up the estimated values from the graph, the Intel chip takes 269 seconds to do each task once. The AMD chip takes 207 seconds.

That's 30% faster!

 

[richierich1212]

The 3770K Rig had 16GB (2 sticks) of DDR3 2300 RAM CAS 16 or 18 timings (I forget exactly, but I bought it in 2012 so it's old)

The 1800X Rig had 16GB (2 sticks) of DDR4 3000 RAM and CAS 14 1T

The thing that really surprised me was 400 Mhz on a 6 year old CPU was significantly faster than any IPC gains made over that same time period, based on my tests, for processes that weren't as multi threaded. My test wasn't perfect but it was a very fair and controlled "real world" test, using the same images and everything, doing exactly what I use it for. Anything heavily multi-threaded though and the Ryzen destroyed it, but that was in line with expectations.

I REALLY want to buy Ryzen

So did you compare with a buddy's system? Or did you build a new Ryzen system? I'm confused. I don't even think Ryzen can run DDR4-3000 at CL14. 2933 or less and/or 3200 and higher.

You may want to double check the memory settings to make sure it wasn't gimped at 2133.

 

[imported_jjj]

Cool. Thanks for the non-snarky response.

The folks around here don't like that site at all as it's not reliable and is less than tech savvy.
What i dislike the most about them is that they present their own speculations as facts. They find a a bit of news and they expand on it at will without noting that it's just their own speculation.
They spot some leaks before others sometimes but other than that it's not a good site.