We need AMD Ryzen Benchmarks! Share your toughts!

[domokun 515]

On 15/04/2017 at 1:14 AM, dwarden said:

 experience of people i know who have it (some i know in person)

Benchmarks seem to suggest otherwise...

 

[Dwarden 1125]

@domokunyou do realize you posting old video (with A3 benchmark), before all the bios/firmware, OS, drivers and Arma 3 updates right ? ;)

[domokun 515]

24 minutes ago, dwarden said:

@domokunyou do realize you posting old video (with A3 benchmark), before all the bios/firmware, OS, drivers and Arma 3 updates right ? ;)

The first video (A3 perf with R7) is indeed a month old but IMHO performance is unlikely to have changed by +/- 10%.

As for the second video (impact of high-speed RAM on R7), i.e. 9 days old.

In neither case, it doesn't strike me as "old" (yes I know 1.68 hit Stable 1 month ago).

But as your is a fair remark, I'll ask the tester to include A3 in his next Ryzen benchmarks.

[Dwarden 1125]

ironically in this very same thread are people with 10 to 20% gains on ryzen already ...

 

the ram testing video is unclear to me

no exact timings, no idea if it runs on the bios which cut of 6ns off the latencies etc.

also benchmarks in 1080 don't matter that much if the bench is GPU bound

plus the issue is those tests are done with NVIDIA GPU only

it would be best to see tests with 480X (580X is too new) or/and R9 Fury

 

also as it was explained in several the depth Ryzen reviews

the NVIDIA DX11 (less DX12) drivers are actually not good for Ryzen due to way how they clutter cpu (overhead)

[Brisse 78]

54 minutes ago, domokun said:

The first video (A3 perf with R7) is indeed a month old but IMHO performance is unlikely to have changed by +/- 10%.

 

My 1700X-system (R9 Fury @1440p and 64GB @2400MT CL14) is currently doing 16-17% better in YAAB compared to when Ryzen launched. That's without any overclocking. The biggest individual gain came after Windows 10 Creators Update. That alone did 9%.

 

Keep in mind that this is a lot higher than the average gains if you look at other games and applications. Arma 3 is actually one of the games which has gained the most from all the OS and BIOS-updates.

[domokun 515]

9 minutes ago, Brisse said:

 

My 1700X-system (R9 Fury @1440p and 64GB @2400MT CL14) is currently doing 16-17% better in YAAB compared to when Ryzen launched. That's without any overclocking. The biggest individual gain came after Windows 10 Creators Update. That alone did 9%.

 

Keep in mind that this is a lot higher than the average gains if you look at other games and applications. Arma 3 is actually one of the games which has gained the most from all the OS and BIOS-updates.

Great news.

I saw the massive Windows Update drop this weekend and suspected as much.

I look forward to more comparative analysis.

[djotacon 190]

I think a lot of video test with arma in Youtube are probably bad or fake.

 

Using this test you can see

 

Ryzen stock vs OC CPU

Ryzen oc 3.9 vs 4.5 - 5.0 OC

None of the shows the distance view

Most of them dont use YAAB

Most of them dont use creators update, game mode.

Most of them dont show minimal fps average

etc, etc ...

 

I waiting a single video of this "awesome" pseudo-IT guys making a comparative with extreme low expensive CPU against top class cpu in Arma 3... I think we see a great surprise in the prize/cpu power segment.

 

[Mahatma Gandhi 29]

7 hours ago, domokun said:

If you compare performance of Intel CPU with AMD Ryzen of comparable price then Intel will deliver better performance in gaming, especially in Arma 3:

@ $250 = i5-7600K vs 1600X

 

I cannot tell much about other games as I am running ArmA3 or I am working with my build (no other games just ArmA3 or rendering, videoediting and encoding). What I can say is that ArmA3 meanwhile runs very very smoothly on my 1700 @ 3.8GHz . I have posted several screenshots of benchmarks and frametimes here. But for you again: ~47 FPS is what I get using standard preset today.

 

 

This is what you would also roughly get with a 1600x @ stock. So instead of posting youtube videos, it would be great if you just run yaab on your build (whatever it is) and post the screenshot here. This will help putting Ryzen into context. Everything else remains pure speculation.

 

Cheers, 

 

Mahatma

[Shaky-AK 0]

9 hours ago, Mahatma Gandhi said:

 

I cannot tell much about other games as I am running ArmA3 or I am working with my build (no other games just ArmA3 or rendering, videoediting and encoding). What I can say is that ArmA3 meanwhile runs very very smoothly on my 1700 @ 3.8GHz . I have posted several screenshots of benchmarks and frametimes here. But for you again: ~47 FPS is what I get using standard preset today.

 

 

This is what you would also roughly get with a 1600x @ stock. So instead of posting youtube videos, it would be great if you just run yaab on your build (whatever it is) and post the screenshot here. This will help putting Ryzen into context. Everything else remains pure speculation.

 

Cheers, 

 

Mahatma

Are you pressing the "S" key when prompted to while running the benchmark?

YAAB shows "standard settings" on the top right corner in gold under "Frames" whenever I end the benchmark with standard settings enabled. (It will also still show "standard settings" in gold when I run the benchmark with standard settings enabled and set a static camera in a safe area away from the action (+384% improvement in fps if I do this). 

 

[Mahatma Gandhi 29]

14 hours ago, Shaky-AK said:

Are you pressing the "S" key when prompted to while running the benchmark?

YAAB shows "standard settings" on the top right corner in gold under "Frames" whenever I end the benchmark with standard settings enabled. (It will also still show "standard settings" in gold when I run the benchmark with standard settings enabled and set a static camera in a safe area away from the action (+384% improvement in fps if I do this). 

 

 

In my list, when standard settings is applied it is mentioned. Standard settings however are not 'standardized' but depend on the preset chosen. Hence, I switched to presets entirely by simply choosing the indicated presets.

 

In other words: If you're running YAAB after selecting ULTRA presets and press 'S' your results will be very different over running YAAB after selecting low preset and hiting the 'S' key.

 

Hence, simply pick standard preset and run YAAB.

[djotacon 190]

I leave this here for explanatory purposes:

 

 

RAM 2400Mhz

[JOHN007 9]

Question is.. What CPU does Bohemia developers use?

[MrHamburgler 0]

I've been eagerly reading this topic here and there over time since I found it from googling, I currently have a ~5 year old i5 3570k with 16gb ddr3 (although currently down to 12gb due to needing to rma a dead stick) with an rx480 on a z77 asrock extreme4m mobo and just doing vanilla a3 missions running around with friends I find my cpu eaten up quite a bit...I also multitask and occasionally stream and wondered if it's worth it to pay the price for maybe the cheaper 12 thread zen with 16gb fast ddr4 3200mhz+ ....does speed or good latency matter more? I do know using the most recent bios and fast ram seems to go best with zen, although a few here have posted good smooth results even from using the cheaper 2400mhz ram. 

 

My fps is decent all things considered, ivy bridge is still pretty good IPC wise, intel really hasn't increased ipc much since sandy bridge back in 2011.....getting an expensive new kaby lake just seems like a horrible value for me tbh vs getting a zen with more threads/cores...I must commend AMD, they went from core2quad era ipc with the sucky per core FX series to haswell-ish level ipc per core with the Ryzens. Apparently the windows10 content creators update improved performance a bit too, and I have heard of a w10 "game mode" recently but I'm still on win7 tbh and don't keep super up to date on windows news. I'm just an average joe gamer but I would like some futureproof-ness in my rig, I find my i5 getting eaten up bad just doing 720p streaming at normal settings. I suppose I could switch to the AMD encoder but many online say that looks way worse than x264 unless you pump the bitrate up more.

 

thinking about upgrading to the 12 thread zen since I think the 16 thread is a bit overkill, but the 1600 seems an amazing value, essentially an expensive high end intel x99 haswell 12 thread cpu for pretty cheap....it would be lovely if cheaper 2400mhz ram would work well....150 bucks for 3200mhz+ ddr4 is a big turnoff tbh, also been thinking of just getting an older i7 used and throwing it into my 1155 mobo. I wouldn't need a new mobo and expensive new ram in addition to the cpu in that case. I know the newer intel are technically still better, but once you consider they're clocked much higher than the new AMD cpu the actual differences would be really small if both clocked to the same speed I'd imagine. What would be a decent mobo without needlessly spending? I'd just use one gpu and overclock a bit to maybe 4ghz, heard the included Wraith cooler is actually very respectable. Also plz no fanboy wars, I'm not a zealot of either company, and have used both in the past, found an 8320E at 4ghz pleasant to use besides Arma and DayZ fps being low, which is why I went back to my older i5 for the much better ipc, but the 4 cores/4 threads just isn't cutting it for what I do. Many say intel didn't "have" to increase their ipc much over the years due to AMD dragging their feet releasing better cpu's, and that may be true, but they shot themselves in the foot tbh since people looking to upgrade from slightly older intel pc's see how the newer intel cpu's are a bad value now, kaby would be what, only about 15% more per core than my ivy bridge? Doubt that'd be noticeable in real world usage...A Zen would still be stronger per core than my ivy bridge.

[Brisse 78]

@MrHamburgler

 

I'm not convinced it's a worthwhile upgrade in your case. Yes, R7 Ryzen's are great for video encoding, but I'm not entirely sure about the streaming argument. If you are an amateur streamer, then AMD VCE is going to be good enough. It's very fast and energy efficient compared to CPU encoding, but it needs about 50% more bitrate to be the same quality, although it depends on your settings and how powerful CPU you are comparing to. You can play around with the settings to improve the quality a bit. You should at least try it for yourself before you dismiss it. If you are a professional, then you most likely already have a separate machine for capture, encode and upload.

[Mahatma Gandhi 29]

@MrHamburgler

 

Well, of course you'll have a huge jump from 4 threads on your i5 to 12 threads on the R5. Also you'll definately have more resources to do something like encoding or streaming in the background but whether this will be a good investment for playing arma totally stands and falls with the RAM you're planing to buy and your current RAM. 

 

We did some benchmarking in a german forum with a guy addressing ARMA3 and RAM performance impact on his overclocked 5GHz i7 7700k and the result is very clear.

 

If your CPU is weak, ARMA performs weak no matter what. However, if your CPU is strong ARMA CAN perform well if your RAM clocks high at short latencies.

 

 

On my 1700 I get ~55FPS @ 3.8GHz with my RAM clocked at 3200MHz (CL14). Using 2400MHz (CL11) RAM he gets to about 47 FPS on a 5GHz i7.

 

 

 

At 3000MHz CL14 and i7 @ 5GHz he is roughly where I am at 3200MHz CL14 and R7@ 3.8GHz suggesting that the R7 performs somewhat better clock by clock (larger caches?). Yet, if he overclocks his RAM to 4200MHz CL17 and i7 @ 5GHz his machine totally takes off. I haven' overclocked my RAM yet as my B350 Board does not offer this option (yet).

 

Most likely this is no different for AMDs Ryzen, it is not advisable to go for a 1700x and run out of money to buy good RAM. This is also true for the i5/i7 obviously. 

 

The sweetspot might be @ 1600x/7600k or if you are up for overclocking, the R5 1600 on a X370 Board is a great choice but under no circumstances go for an expensive CPU at the expense that you'll end up with slow RAM becuase you're running out of money (if you love playing ArmA3).

 

I'd recommend the R5 1600(x) plus TridentZ @3200MHz CL15 (that one has the SamsungB Dice - dont take the CL16! - those are Samsung E and Ryzen does not like those) or FlareX or good DualRank DDR4 that runs the 2667-2993 MHz on a Ryzen Board. It just feels wrong to spend the same amount of money for a QuadCore i5 if you can get 6C/12T.

 

Hope that helps,

 

Mahatma

 

 

P.S.: In my opinion it is worth noting that all these data are very interesting from a theoretical point of view, however, an R5 1600 or i5 7600 with decent RAM are both well within very solid performance in ArmA3 when combined with decent RAM. You don't need >70 FPS in the YAAB to make playing ArmA3 a great experience.

[I'mJimFuckingSterlingSon 10]

i think i will give my i5 4460 to my mom and build myself a ryzen sextop. 

 

 

[+Halina+ 7]

So, I figure it is recomended to update BIOS firmware to improve AMD CPU performance?

[Mahatma Gandhi 29]

6 hours ago, +Halina+ said:

So, I figure it is recomended to update BIOS firmware to improve AMD CPU performance?

 

The latest BIOS/UEFI update (AGESA 1.0.0.4a) gave shorter latencies accessing the RAM.

[Fiddich 0]

Like it or not you can't blame AMD. This falls solely at the feet of BIS for making a product that works better on one hardware over another.  Rest assured AMD did not likely give two shits how well one poorly coded game that has a minute market share works on their chips.  We as consumers should demand more from the software devs to write better fracking code.  Stop giving them money, stop playing their games, and just maybe they will make something of quality.  Otherwise build your machine based on the consensus of what works best.  I say this having just finished building an AMD Ryzen 5 set up. And did so knowing full well that Arma and BIS have no idea how to code for multi core processors....I mean really, it's 2017....and how long have multi core processors been around...but I digress.

[focher 15]

5 hours ago, Fiddich said:

Like it or not you can't blame AMD. This falls solely at the feet of BIS for making a product that works better on one hardware over another.  Rest assured AMD did not likely give two shits how well one poorly coded game that has a minute market share works on their chips.  We as consumers should demand more from the software devs to write better fracking code.  Stop giving them money, stop playing their games, and just maybe they will make something of quality.  Otherwise build your machine based on the consensus of what works best.  I say this having just finished building an AMD Ryzen 5 set up. And did so knowing full well that Arma and BIS have no idea how to code for multi core processors....I mean really, it's 2017....and how long have multi core processors been around...but I digress.

 

You can't digress when you've never posted before. Seriously, you registered and made your this your first contribution to the community?

[djotacon 190]

1 hour ago, focher said:

 

You can't digress when you've never posted before. Seriously, you registered and made your this your first contribution to the community?

 

This is only an opinion like others not a contribution, everybody has the right to post his own opinions.

 

I dont understand why someone is judged by his "contribution" to the community and I think that the comment of fiddich is very smart and "nails" the point of this debate.

 

In the other hands the title of the thread is "share your thougts".. and I believe this thread dont need any mandatory community contribution.

 

mmm ... I try to understand this : " You can't digress when you've never posted before "... who say that? you? but the comment of fiddich is related with the thread, I dont see any offtopic here, I believe this kind of opinion is a labor of the forum admins.

 

I dont understand too why some members of this community thinks a valuable member is someone that need mandatory "contribution", no one in this community needs to do nothing for anyone all in this community - as I know - is released for free.

[Fiddich 0]

Djotacon- thanks.  

 

I came here solely to learn if there were any tweaks i can do to make the performance better.  Over and over and over again most "regular" contributors state "well you shoulda bought intel" or some other fanboy nonsense.  I have for 15+ years and will likely always prefer AMD.  Call me a fanboy or whatever but I just prefer AMD.  

 

Like said, this is 100,000 percent Bohemia's fault, problem, laziness, whatever.  It's not really that much different than say Gran Turismo ONLY being made for the Sony hardware. It's up to the Dev to make it run on an X-Box, but they don't. It's not up to MS make that game run on an X-box. 

 

Not enough "contributors" hold the devs hand to the fire to make a product that works properly.  If this were really true games like BF and COD would have ruined those pubs/devs. But the consumer just doesn't do enough to make these businesses account for their incomplete products and just allows for them to release a "patch". Yeah I'm an old fart who grew up on games that never got patched cause they mostly never needed it. Now it's just biz as usual for this practice. /rant. 

 

Anyway look forward to seeing more results from other AMD fanboys. 

[djotacon 190]

12 hours ago, Fiddich said:

Djotacon- thanks.  

 

fanboys. 

 

This  word is an insult to everyone on internet and in this forum is against forum rules too - you can use AMD users, AMD buyer, AMD whathever-, please stop using this word.

[Shaky-AK 0]

Quote

Like it or not you can't blame AMD. This falls solely at the feet of BIS for making a product that works better on one hardware over another.  Rest assured AMD did not likely give two shits how well one poorly coded game that has a minute market share works on their chips.  We as consumers should demand more from the software devs to write better fracking code.  Stop giving them money, stop playing their games, and just maybe they will make something of quality.  Otherwise build your machine based on the consensus of what works best.  I say this having just finished building an AMD Ryzen 5 set up. And did so knowing full well that Arma and BIS have no idea how to code for multi core processors....I mean really, it's 2017....and how long have multi core processors been around...but I digress.

 

 

Why blame AMD, or BIS for anything? People in this thread already confirmed that arma 3 runs well on ryzen.

 

[Fiddich 0]

Seriously...fan boy is an offence word/label here?  yikes.  I've never seen it as an offence label but that's a whole other conversation 

 

Anyways, on to Amra3.

I'm running a Ryzen 1500X on a GB 370X mobo with 16gb of 2100 DDR4 ram and a 290x vid card.

I see anywhere from 30 to 40 fps on thw Arma3 Exile MP server I play on.  much better than the 10 to 15 I got with my 8150 cpu.