We need AMD Ryzen Benchmarks! Share your toughts!

[RustyAK 0]

All the benchmarks I've seen so far have been done with without yaab, or with comedically low viewing distance. I'm hoping someone with an 1800x posts some decent benchmarks before I buy parts, because that would help me choose my build. 

 

P.S. Thanks to whoever posts 1800x a3 benchmarks.

[Mahatma Gandhi 29]

On 3/28/2017 at 5:20 PM, RustyAK said:

P.S. Thanks to whoever posts 1800x a3 benchmarks.

 

I can give you a YAAB bench on an R7 1700 @3.8GHz + RX470 if you tell me which settings you would like to see benched.

 

But to what I have experienced with this System, FPS is not all. It runs very smooth even if I run into low FPS scenes. I am not kidding, it really is a much smoother feeling.

[RustyAK 0]

1 hour ago, Mahatma Gandhi said:

 

I can give you a YAAB bench on an R7 1700 @3.8GHz + RX470 if you tell me which settings you would like to see benched.

 

But to what I have experienced with this System, FPS is not all. It runs very smooth even if I run into low FPS scenes. I am not kidding, it really is a much smoot

1 hour ago, Mahatma Gandhi said:

 

 

 

Thanks Mahatma yaab @ both standard settings, and another one at max viewing/object distance (other graphics settings aren't a big deal to me) 1080p, or higher. That would pretty much help me choose between intel, and amd for a cpu.

 

[Valken 622]

I want to see Ultra settings and what is your RAM speed set at? Indication from other websites says Ryzen OC'ed @ ~ 4GHZ with 3200 CL14, 3600 CL14 or faster ram goes 1:1 with an Intel 7700K @ 5GHZ with 3200 CL14 RAM.

 

If Ryzen can get that high performance out of ARMA, I will pass my current Haswell @ 4.4 GHZ to my kids and get a Ryzen 1700 for myself when I upgrade the GPU. 

 

 

[Mahatma Gandhi 29]

1 hour ago, Valken said:

I want to see Ultra settings and what is your RAM speed set at? Indication from other websites says Ryzen OC'ed @ ~ 4GHZ with 3200 CL14, 3600 CL14 or faster ram goes 1:1 with an Intel 7700K @ 5GHZ with 3200 CL14 RAM.

 

If Ryzen can get that high performance out of ARMA, I will pass my current Haswell @ 4.4 GHZ to my kids and get a Ryzen 1700 for myself when I upgrade the GPU. 

 

 

Are you using an i5 or i7 Hasswell?

 

I am happy to bench what ever is interesting to you guys but as a remark, an RX470 is not exactly a 1080Ti. Dont expect wonders under Ultra Settings (which include 8xAA) ;-)

 

I can also walk to a predetermined position in the build in showcases so you can compare in game (SP)

 

Will do it later when I am back from work.

 

[djotacon 190]

I this is offtopic but relevant for the future of improvements related with the RYZEN platorm:

 

[Mahatma Gandhi 29]

Ok, so under 3.8GHz the YAAB results in 36.8 FPS (DDR4 @ 3200 MHz)

 

under total standard settings (RAM at 2400MHz CL16) and CPU at 3.2GHz - 3.7GHz Boost its 29.4 FPS and finally, STD settings and DDR4 at 3200MHz it is 33.3 FPS

 

The RX470 was at std at all times

[djotacon 190]

35 minutes ago, Mahatma Gandhi said:

Ok, so under 3.8GHz the YAAB results in 36.8 FPS (DDR4 @ 3200 MHz)

 

under total standard settings (RAM at 2400MHz CL16) and CPU at 3.2GHz - 3.7GHz Boost its 29.4 FPS and finally, STD settings and DDR4 at 3200MHz it is 33.3 FPS

 

The RX470 was at std at all times

 

You cannot use mods in a test because is a test for rendering speed with a mission, with a mod you lost more fps your test rigth now with mods is invalid.

 

And you have more fps enabling the high power comsumption profile in the o.s. energy options to avoid coreparking.

 

To make a proper measure change the distance view to a urban warfare - 500m- more than that you have severe overdraw because buildings, - that why there's 300/500m in KOTH mission-

 

[Valken 622]

I am using i7-4790K @ 4.4 GHZ with DDR3-2400 CL10 RAM. AMD 6950 unlocked in CFX. I will do a quick run when I return from work as Steam is still updating.

 

Make sure to unload all MODs as it does affect performance but your FPS is real good since with MODs, I get down to 16 FPS (ACE, BLASTCORE and a few more) :D

 

You can run ULTRA and turn off AA. Just use FXAA or SMAA.

[RustyAK 0]

4 hours ago, Mahatma Gandhi said:

Ok, so under 3.8GHz the YAAB results in 36.8 FPS (DDR4 @ 3200 MHz)

 

under total standard settings (RAM at 2400MHz CL16) and CPU at 3.2GHz - 3.7GHz Boost its 29.4 FPS and finally, STD settings and DDR4 at 3200MHz it is 33.3 FPS

 

The RX470 was at std at all times

Good stuff. Again thanks for taking the time it really is an eye opener. My current pc tops out at 21-22 fps on standard with, or without mods enabled.

[Mahatma Gandhi 29]

20 hours ago, djotacon said:

 

You cannot use mods in a test because is a test for rendering speed with a mission, with a mod you lost more fps your test rigth now with mods is invalid.

 

And you have more fps enabling the high power comsumption profile in the o.s. energy options to avoid coreparking.

 

To make a proper measure change the distance view to a urban warfare - 500m- more than that you have severe overdraw because buildings, - that why there's 300/500m in KOTH mission-

 

 

I don't think that the CBA_A3 Mod has much of an impact here but technically you are of course right. You can maybe add an extra frame*s^-1 but again, its not making a b/w difference. The sys was under 'High Performance' windows settings, so core parking should have been disabled in this run (I did not doublecheck). HPET was enabled, as I need it for some other stuff I am doing. Beside the YAAB I can definitely report excellent performance and a very smooth game play. My in game settings always are ULTRA preset with maxed out view distance (12,000m) in FHD.

 

If I would look for a general purpose rick which needs to perform well under ArmA3 I would probably choose between the upcoming Ryzen 5 1600x with 6c/12T or if the budget allows the 1700x (8c/16T) depending on which other tasks you're running on the sys. Me for example, I am often running an Skype video conference with old friends on my second screen if we do multiplayer and thats where you really dont have to worry anymore with the big ryzen sys. Definitely, I would go for the 1600x if my budget does not allow a 1700x AND fast RAM. Ryzen really likes fast RAM and ArmA3 does also. If you go for 1700x + DDR4 2133MHz you will be disappointed. For the same money, you'll have a 1600x + G.Skill FlareX 3200MHz CL14 which is probably the sweetspot for the majority of the people.

 

@ djotacon I also see the potential that the Ryzen 'ages well' as more and more games will get threaded and or see optimization towards the Ryzen architecture. According to Oxide Games these optimizations that now led to >20% gain on the Ryzen µArch took them 400 man hours. Sounds a lot but is basically 3-4 people working full time for 3 weeks. But realistically, and Dwarden you might forgive me, it took BIS many many years to adopt the ArmA engine to AMD64 and I do not have the feeling they are going to invest even 100 manhours (please proof me wrong!) to optimze towards a specific architecture. I do think however, that this would be a great opportunity for BIS as the Ryzen µArch is brand new, sells fantastic, is already fast in ArmA3 and will stay for many years to come. Right now, BIS might even get very good assistance from AMD if optimizing towards their new µArch as it is also in their own interest. But again, I am not holding breath given how long it took to release an AMD64 version of the ArmA engine.

 

Best, Mahatma

 

EDIT: Here are the Results using the Ultra Preset + all maxed out

 

 

 

[Horus 83]

AMD Fixes More Ryzen Issues with New BIOS Firmware Microcode ( early April ) http://www.guru3d.com/news-story/amd-fixes-more-ryzen-issues-with-new-microcode.html

Quote

 

As a brief primer, the AGESA is responsible for initializing AMD x86-64 processors during boot time, acting as something of a “nucleus” for the BIOS updates you receive for your motherboard. Motherboard vendors take the baseline capabilities of our AGESA releases and build on that infrastructure to create the files you download and flash.

We will soon be distributing AGESA point release 1.0.0.4 to our motherboard partners. We expect BIOSes based on this AGESA to start hitting the public in early April, though specific dates will depend on the schedules and QA practices of your motherboard vendor.

BIOSes based on this new code will have four important improvements for you

1. We have reduced DRAM latency by approximately 6ns. This can result in higher performance for latency-sensitive applications.
2. We resolved a condition where an unusual FMA3 code sequence could cause a system hang.
3. We resolved the “overclock sleep bug” where an incorrect CPU frequency could be reported after resuming from S3 sleep.
4. AMD Ryzen™ Master no longer requires the High-Precision Event Timer (HPET).

The new BIOS updates with updated micro-code will be released in April. Below you can check some benchmark results (from AMD) on Ashes of the Singularity and DOTA 2.

 

 

[Dwarden 1125]

well let's hope Ryzen and mainboards can run well with low latency DDR4 modules at 3200MHz (or eventually 3733/4266)

then those 6ns (i hope more is squeezed in future) can come handy 

 

i wonder what the 4. means exactly like no more issue with hpet on/off or else

[Mahatma Gandhi 29]

On 4/2/2017 at 2:57 AM, dwarden said:

i wonder what the 4. means exactly like no more issue with hpet on/off or else

 

HPET has always been in debate to impact performance in some games, thats not a Ryzen specific thing. So AMD suggests to disable it. Yet, their own Software 'Ryzen Master' requires it to be enabled. The latter is going to change.

[Mahatma Gandhi 29]

After my Board got a new Bios, I benched a little more using default presets. 

 

 

I think it looks very good!

 

One important thing I figured out is that YAAB results are biased based on which preset you're coming from when you apply standard settings. Thats why i got much higher scores in this run. So if you want to compare, use the presets not the standard settings.

[St. Jimmy 272]

1 hour ago, Mahatma Gandhi said:

After my Board got a new Bios, I benched a little more using default presets.

 

I think it looks very good!

Looks like it's sitting just behind level of i7 2600K @ 4,7GHz and CL9 2133MHz RAM. Here's a comparsion:

I did the same thing as you, change the setting to Ultra and then press S for standard settings at the beginning of the benchmark. Too bad that my adaptive vsync didn't catch that 72fps spike down to get better comparison :/

[Mahatma Gandhi 29]

@ St. Jimmy

 

Pretty close for almost a GHz in between ;-)

 

Whats your GPU?

[St. Jimmy 272]

3 hours ago, Mahatma Gandhi said:

@ St. Jimmy

 

Pretty close for almost a GHz in between ;-)

 

Whats your GPU?

Yep pretty close, though it's over 5 years old CPU. Too bad that the Ryzen doesn't clock much further :(

I've GTX 1070. I just realized to read that you've only RX470. Do you hit GPU bottleneck at any time in the benchmark?

 

//Quickly tested with low preset and I got 56fps, so I guess the GPU doesn't really matter.

[Mahatma Gandhi 29]

On 4/6/2017 at 3:27 PM, St. Jimmy said:

Yep pretty close, though it's over 5 years old CPU. Too bad that the Ryzen doesn't clock much further :(

 

Well, you know... Playing ArmA3 is not the only thing I am doing with this rig and it does somewhat better than your 2600k in other applications.

 

 

And thats actually representative if I work on it with Imaris, ZEN.... to render stuff.

 

EDIT: here are the frametimes during YAAB (standard preset + standard settings) if somebody is interested: (This is actually everything between the plane dropping bombs in the beginning and "benchmark ended")

 

The highest percent of frametimes is averaging at 34.1ms

The highest five percent of frametimes is averaging at 30.3ms

 

Now beat me St. Jimmy ;-)

[djotacon 190]

There's a new energy profile for windows 10 dedicated to increase performance:

 

https://community.amd.com/community/gaming/blog/2017/04/06/amd-ryzen-community-update-3

[Brisse 78]

I installed Windows 10 Creators Update today, and my YAAB-result went up 9% on my 1700X-machine :)

 

Not sure if it's some Ryzen optimization, Game Mode or anything else.

[Moerderhoschi 107]

not ryzen related but a nice info of the RAM Mhz impact on arma frames:

 

source: http://techreport.com/review/31179/intel-core-i7-7700k-kaby-lake-cpu-reviewed/11

 

[Brisse 78]

BIOS updated for my PRIME X370-PRO and 1700X. Version number is now 0604 and this update contains the new AGESA from AMD which promises important fixes and better performance.

 

Windows 10 Creators Update previously gave me a 9% boost in YAAB, and this new BIOS gave me another 4% better performance over my previous run.

 

Overall, I'm seeing about 15% better results in YAAB since Ryzen launch. Pretty good if I may say so!

[Housesparrow 8]

I am still waiting for my G.Skill Flare X, CL14 3200MHz ramsticks to even ship. Estimated time to it shipping being 14+ days. :( Looking forward to test with YAAB now that I found YAAB on Steam.

 

Below, my incomplete Ryzen build:

• 1800x cpu
• Nofan CR-95C passive cpu cooler (AM3)
• G.Skill Flare X 2x8GB, CL14, 3200MHz ram
• Asus Rog Crosshair VI Hero board (AM4/AM3)
• Asus Rog Rx480 Strix 8GB OC gfx card 

• Samsung 960 Evo SSD (M.2 socket)
• Win 10 pro on a usb stick (only to be used for gaming)
• Seasonic fanless psu 520W
• Cablemod's white basic cable pack for Seasonic FL2 psu
• Fractal Design Define S w. window

[bingo_fuel 10]

On 11.4.2017 at 2:26 PM, Brisse said:

BIOS updated for my PRIME X370-PRO and 1700X. Version number is now 0604 and this update contains the new AGESA from AMD which promises important fixes and better performance.

 

Windows 10 Creators Update previously gave me a 9% boost in YAAB, and this new BIOS gave me another 4% better performance over my previous run.

 

Overall, I'm seeing about 15% better results in YAAB since Ryzen launch. Pretty good if I may say so!

 

A 15% improvement within a few weeks is really impressive!