The 2019 "Put your Benchmark where your Mouth is" thread.

[sawdust3d]

Hey All,

It's an exciting time, especially with all the great CPU releases. We get a lot of speculation and 3rd hand info batted around here in the Steam Hardware and OS community: lets see what real world users are getting out of the near unlimited hardware combinations and implementation options.

We're using the 3DMark suite here for testing as it covers all the currently relevant resolutions and is uniformly strict in validating scores. I, like many of you probably have OCs that crush in game & under stress testing but can't pass some 3DMark benches...tough..drop your OC or up your voltage, that's why we're using it.




Here's a simple format for Copy/Paste:

CPU: OCd: Y/N
GPU: OCd: Y/N

Cooling Solution: Air/Water/Combined (AIO)/Sub-Ambient

Benchmark Scores:

FS (Fire Stike):............................
FSE (Fire Strike Extreme):..........
FSU (Fire Stike Ultra):.................
TS (Time Spy):............................
TSE (Time Spy Extreme):...........




Rules/Guidelines:
1- No Pick = No Proof (All scores must be accompanied by supporting evidence)
2- If you are not participating and posting scores; please post you commentary, analysis or questions in a separate thread.
3- Keep it nice; a little friendly competition is fun, but we're not gonna have any disparaging remarks, fanboyism or "mine is better than yours" BS. We have an opportunity here for a cool base of knowledge.
4- This is not for "UBER" machines only. While we're all interested to see about the 3000 series vs the 9000 series, that is not the only point. If you've got your hamsters all fired up on caffiene & the O'le potato is putting out 1.21 Gigawatts then rock on! Let's see those scores.


I'll put up my stock score as baseline example & we'll see from there.

OC it till it squeaks! - Good Luck!

[SHREDDER]

AbedsBrother님이 먼저 게시:

SHREDDER님이 먼저 게시:

Is the latest verson of 3d mark free? Ryzen 7 1700 16 GB DDR4 3200MHZCL15 GS KILL RIP JAWS GTX 970 G1 GAMING DELL2P416D 24'' 2560X1440 60 HZ IPS. Wait for my results! IAM THE SHREDDER AND WILL BEAT YOU ALL!

I challenge you to beat my Time Spy score (3987)!

The RX 470 and GTX 970 are fairly close - curious to see how close. I expect the GTX 970 to score around 4400 or so.

Did you ran the basic version?

[AbedsBrother]

SHREDDER님이 먼저 게시:

AbedsBrother님이 먼저 게시:

I challenge you to beat my Time Spy score (3987)!

The RX 470 and GTX 970 are fairly close - curious to see how close. I expect the GTX 970 to score around 4400 or so.

Did you ran the basic version?

Demo + Bench. Wasn't aware there was any other version of Time Spy, even in the free version of the software.

SHREDDER님이 먼저 게시:

AbedsBrother님이 먼저 게시:

I challenge you to beat my Time Spy score (3987)!

The RX 470 and GTX 970 are fairly close - curious to see how close. I expect the GTX 970 to score around 4400 or so.

It is GTX 9 70 G1 GAMING which gtx 970 do you have?

I have a Radeon RX 470.

SHREDDER님이 먼저 게시:

Last time i run it was in 2017 when i upgraded cpu and ram(from core i5 2500k and 12 gb ddr3 1600 mhz to ryzen 7 1700 and 16 gb ddr4 3200mhzcl15) but i dont remeber the score number. Do you also have a 1440p monitor like me? What cpu and ram do you have? Iam running it each time i upgrade so i will run it again when i upgrade graphics card and i believe then it will be much higher than 20.000 especialy if the card is 50%+ more powerful than RTX 2080 ti.

I have:
Ryzen 7 2700 - OC'd to 3.6Ghz all cores @1.150V
It could go higher, but I'm using the stock cooler and have a high ambient temp - around 27C.
Asrock B450M Pro4
16GB G.Skill Flare-X 3200Mhz cas16
Radeon RX 470 (4GB MSI Armor) +50% power limit
1 TB Adata Su800 SSD for (OS & games)
1TB Western Digital Blue 7200rpm (for recording & editing video footage)
Rosewill ARC-M650
My monitor is a 27", 1080p Acer, 75hz w/ Freesync.

[AbedsBrother]

Dr Doom님이 먼저 게시:

Games would be done by testing a wide range, this would include games which are CPU bound and GPU bound. From this data based on frame-varience, frame-rate lows, highs and averages by percentage. Over time this gives you a clear look at overall gaming performance; and I need to correct myself IPC gains can be significant over time generation to generation. However when you compare some proessors. There is a clear distinction between how much of a gain there is if you lock their clocks at 4Gh or some other absolute value. Then by comparing them side by side in terms of the frame-rate data you get a deeper understanding of that processors real performance.

Productivity would be using: 3D rendering software, CPU bound fluid simulation, AutoCAD, or other benchmarks.


To try to determine real world performance from synthetics tells you little about how varied workloads behave when applied to a system.

I understand what you're saying here - it's why when I benchmark a game, I make an hour-long video of unedited gameplay - but I think there is still a place for theoretical benchmarks. They are baseline of performance tests that make the exact same demands of every machine that runs them.

[bigbenisdaman]

AbedsBrother님이 먼저 게시:

Dr Doom님이 먼저 게시:

Games would be done by testing a wide range, this would include games which are CPU bound and GPU bound. From this data based on frame-varience, frame-rate lows, highs and averages by percentage. Over time this gives you a clear look at overall gaming performance; and I need to correct myself IPC gains can be significant over time generation to generation. However when you compare some proessors. There is a clear distinction between how much of a gain there is if you lock their clocks at 4Gh or some other absolute value. Then by comparing them side by side in terms of the frame-rate data you get a deeper understanding of that processors real performance.

Productivity would be using: 3D rendering software, CPU bound fluid simulation, AutoCAD, or other benchmarks.


To try to determine real world performance from synthetics tells you little about how varied workloads behave when applied to a system.

I understand what you're saying here - it's why when I benchmark a game, I make an hour-long video of unedited gameplay - but I think there is still a place for theoretical benchmarks. They are baseline of performance tests that make the exact same demands of every machine that runs them.

WOw, hour long vids...yea i'd never watch lol.

Benchmarks like this is good for quick comparisons. Sure some brands perform better than others in certain games, but for someone buying new hardware a quick look will get them where they want to go. If you only play 1-2 games, check htem benchmarks out.

Also benchmarks are highly consistant, meaning if you want to test performance they're a good way to determine if your changes mean anything.

Even an hour long vid of gameplay will smooth over any hitches/stutters soem people experiance.

[Ad Hominem]

Made it back from vacation and ran these on my main desktop. For some reason I was not able to make time spy run, it crashed as soon as loading began. Going to have to try verifying files and whatnot, but I have a lot of unpacking to do first.

i7 8700k
1080ti FTW 3
16GB Tridentz 3200 CL14 RAM
MSI Z370 Gaming M5
Samsung 970 EVO 500GB boot drive


CPU: OCd: Y - 5.0GHz
GPU: OCd: Y - +75/500

Cooling Solution: Air (NH-D15 on CPU)

Benchmark Scores:

FS (Fire Stike): 23,172
FSE (Fire Strike Extreme): 13,583
FSU (Fire Stike Ultra): 7,366



https://steamcommunity.com/sharedfiles/filedetails/?id=1809184205

https://steamcommunity.com/sharedfiles/filedetails/?id=1809184971

https://steamcommunity.com/sharedfiles/filedetails/?id=1809185421

[DevaVictrix]

i3 6100
R9 380 4GB

Both at stock speeds using stock cooling.

Benched on Ubuntu 19.04 using Steam Play

Fire Strike 6004
(Graphics score 7286, 36.24fps, 28.14fps)
(Physics score 6236, 19.8fps)
(Combine 2529, 11.86fps)

I can't run any of the other benchmarks... apparently I need DX12.

[bigbenisdaman]

DevaVictrix님이 먼저 게시:

i3 6100
R9 380 4GB

Both at stock speeds using stock cooling.

Benched on Ubuntu 19.04 using Steam Play

Fire Strike 6004
(Graphics score 7286, 36.24fps, 28.14fps)
(Physics score 6236, 19.8fps)
(Combine 2529, 11.86fps)

I can't run any of the other benchmarks... apparently I need DX12.

Try win 10 and I think your gpu supports dx12.

[Omega]

bigbenisdaman님이 먼저 게시:

DevaVictrix님이 먼저 게시:

i3 6100
R9 380 4GB

Both at stock speeds using stock cooling.

Benched on Ubuntu 19.04 using Steam Play

Fire Strike 6004
(Graphics score 7286, 36.24fps, 28.14fps)
(Physics score 6236, 19.8fps)
(Combine 2529, 11.86fps)

I can't run any of the other benchmarks... apparently I need DX12.

Try win 10 and I think your gpu supports dx12.

The benchmarking tool is just stupid, on Linux we translate DirectX 10/11/12 over to Vulkan to play our games. If the benchmark allowed us to just start it it would work just fine. The benchmark is programmed to look for DirectX12 and remove the start button if it's not found.

This tech is becoming so good now that many Windows games are actually running better on Linux then natively on Windows.

[r.linder]

SHREDDER님이 먼저 게시:

Escorve님이 먼저 게시:

Basic version is free but only comes with Time Spy. And you won't beat us, because the best score for your config hasn't even reached 5000 in standard Time Spy, which is nothing compared to the top scores of over 30000.

Those top scores are only with 2 RTX 2080Ti's and 12+ cores cpus which now became afffordable with Ryzen 3900x. Most people dont and will not pay so much money just to post those crazy high scores where there is not realy reason to own such expensive hardware.

3900X is slower than a 9900K in 3DMark last I checked. And the top score is with a 9980XE and 4 1080 Tis, not even 3950X would top that.

[Ad Hominem]

Escorve님이 먼저 게시:

SHREDDER님이 먼저 게시:

Those top scores are only with 2 RTX 2080Ti's and 12+ cores cpus which now became afffordable with Ryzen 3900x. Most people dont and will not pay so much money just to post those crazy high scores where there is not realy reason to own such expensive hardware.

3900X is slower than a 9900K in 3DMark last I checked.

It depends on how parallelizable the particular task at hand is.

[bigbenisdaman]

Omega님이 먼저 게시:

bigbenisdaman님이 먼저 게시:

Try win 10 and I think your gpu supports dx12.

The benchmark is just stupid, on Linux we translate DirectX 10/11/12 over to Vulkan to play our games. If the benchmark allowed us to just start it it would work just fine.

This tech is becoming so good now that many Windows games are actually running better on Linux then natively on Windows.

Idk anything bout linux, but would have to see some reviews on it running better than windows going through a conversion thingie. It may have better fps, but lack features etc.

[Omega]

bigbenisdaman님이 먼저 게시:

Omega님이 먼저 게시:

The benchmark is just stupid, on Linux we translate DirectX 10/11/12 over to Vulkan to play our games. If the benchmark allowed us to just start it it would work just fine.

This tech is becoming so good now that many Windows games are actually running better on Linux then natively on Windows.

Idk anything bout linux, but would have to see some reviews on it running better than windows going through a conversion thingie. It may have better fps, but lack features etc.

Your inexperience with Linux is obvious from your statement. People run Linux for the features, it can do a lot more then Windows can, it can very easily be interacted with and is highly customizable.

Your router runs Linux, your Android phone also runs Linux even your smart refrigerator runs Linux. Most websites you visit are hosted on Linux, pretty much everything except for the desktop is dominated by Linux.

The Playstation 4? That runs FreeBSD, a distant cousin of Linux. They both share a lot of code.

People used to avoid it because it could barely game. Now it can game and it can do it a really well.

[bigbenisdaman]

Omega님이 먼저 게시:

bigbenisdaman님이 먼저 게시:

Idk anything bout linux, but would have to see some reviews on it running better than windows going through a conversion thingie. It may have better fps, but lack features etc.

Your inexperience with Linux is obvious from your statement.





People run Linux for the features, it can do a lot more then Windows can, it can very easily be interacted with and is highly customizable.

Your router runs Linux, your Android phone also runs Linux even your smart refrigerator runs Linux. Most websites you visit are hosted on Linux, pretty much everything except for the desktop is dominated by Linux.

The Playstation 4? That runs FreeBSD, a distant cousin of Linux. They both share a lot of code.

People used to avoid it because it could barely game. Now it can game and it can do it a really well.

yea, i literally said so lol.

Run whatever you want, maybe linux will have dx12 support one day (without missing features).

An nah, i don't play on console, just pc nowadays.

Linux still lacks support for many titles, 3d mark for example....

An no, my phone/fridge etc don't run linux. My router may, don't know, don't really care.

[DevaVictrix]

I dunno. I feel I take a bit of a performance hit for using Linux. I've just been doing some googling... In Windows with DX12 I should have been getting 7000+. Who knows using DX11.

Rise of the Tomb Raider (ported) played better than many Youtube videos, Mad Max was about on par with WIndows but GTA 5 through Steam Play was far worse. I think my i3 threw a spanner in the works with that game. On Windows crossfiring two R9 380s didnt help fps at all.
The difference of course is DX and Vulkan. Linux doesn't have DX so has to emulate. Vulkan (in my slim knowledge of it all) is better!

I've hated Linux vs Windows discussions ever since I first heard of Linux (>20 years ago) but I'm OK with the performance hit using Linux. It's getting sooo much better recently but i'll agree with the crowd that says "WIndows for gaming, Linux for everything else".

[bigbenisdaman]

DevaVictrix님이 먼저 게시:

I dunno. I think my i3 threw a spanner in the works with that game. On Windows crossfiring two R9 380s didnt help fps at all.

.

Nah,that's just GTA and it's heavy CPU needs. crossfiring them with an I3 is kinda insane for that game lol. Even with the best cpu's gta peaks out 1-2 cores if fps is uncapped with just a 1070.

An yea, fanboys are dumb. No matter what it is, intel/ amd. Nvidia/amd. Windows/linux. If it does what you want it to, so be it. Though even now I can't ever see running stictly linux due to lack of support.