PPL - Performance Per Liter, do you SFF?

[Duality92]

Round 1 is done go HERE for round 2!​

General rules and entery rules

• Picture of actual PC
• Volume calculated with exterior dimensions of case WITH power brick(s)
• CPU-Z / GPU-Z screenshots obligatory, see post #6
• CPU-Z must be run on "Benchmark" and validation link saved with your username
• GPU-Z must be on the first tab when doing screenshot with a dedicated GPU
• Heaven Benchmark on "Extreme" preset
• PPL calculation ((CPU-Z Multi Thread score * Heaven Benchmark score)/Build Volume)/1000
• This is a friendly competition, keep it clean and civil
• Competition ends at 25 entries (new round with new benches starts after)

Leaderboard (PPL - Username - CPU - GPU - (CPUZMT*Heaven)/Volume - Post #

1. 7179 - Gautam - i9 7960X - GTX 1080 Ti - (11537*4916)/7.9 - Post #232
2. 3732 - Arie - i9 7980XE - GTX 1080 Ti - (9776*4390)/11.5 - Post #133
3. 3381 - petricor - i7 8700k - GTX 1080 - (4349*3887)/5.0 - Post #186
4. 3075 - Damascus - i7 6950X - GTX 1080 Ti - (5053*4443)/7.31 - Post #169
5. 2579 - n4ru - i7 8809G - RX Vega M GH - (2930*1426)/1.62 - Post #181
6. 2309 - GermanDrifter97 - R7 1700 - GTX 1080 - (4574.9*3382)/6.7 - Post #102
7. 2298 - confusis - i7 8700 - GTX 1060 6GB - (3879.7*2073)/3.5 - Post #121
8. 2246 - TroyMSLprod - i7 8700K - GTX 1070 - (4423.6*2975)/5.86 - Post #285**
9. 2140 - rfarmer - i7 8700K - GTX 1070 - (4520*3092)/6.7 - Post #74
10. 1969 - ZMan - i7 7700K - GTX 1070 - (2533*2938)/3.78 - Post #114
11. 1670 - Rysen - i7 7700K - GTX 1080 Ti - (2620.7*4589)/7.2 - Post #185
12. 1481 - Testifier - i7 7800X - GTX 1080 Ti SLI - (3297.8*4887)/10.88 - #33
13. 1206 - Freeks - R5 1600 - GTX 1080 - (3151.7*3100)/8,1 - Post #48
14. 1185 - TheHIg - R5 2600 - GTX 1060 6GB - (3388*2050)/5.86 - Post #56
15. 1153 - firewolfy - i7 6700K - GTX 1070 - (2655*2910)/6.7 - Post #57
16. 1046 - Duality92 - i5 8600K - GTX 1070 - (3669.6*2970)/10.4 - Post #280
17. 984 - axm - i7 6700K - GTX 1080 Ti - (2585.3*4795)/12.6
18. 887 - theGryphon - R7 1800X - GTX 1050 Ti - (4749.4*1232)/6.6 - Post #26
19. 820 - TinyAudio - i7 6770HQ - Intel Iris Pro Graphics 580 - (1890*317)/0.73 - Post #275
20. 562 - Curiosity - i5 3570k - GTX 1060 6GB - (1441.7*1598)/4.11 - Post #7
21. 411 - cstein08 - i5 2320 - GTX 970 - (1211*1537)/4.53 - Post #51
22. 403 - Frazhna - i5 4690 - GTX 1050 2G - (1471.9*1092)/3.99 - Post #81
23. 401 - Duality92 - i5 7300U ES - Intel HD Graphics 620 - (1051.6*240)/0.63 - Post #6
24. 367 - VegetableStu - i7 6800K - GTX 1060 6GB - (3383.4*2134)/19.65 - Post #125
25. 212 - jmarin - R5 2400G - GTX 1050 Ti - (2399*1181)/13.37 - Post #37
26. 190- TinyAudio - i7 7700T - Intel HD Graphics 630 - (2146*195)/2.2 - Post #282

** Score #8 was #26 entry, technically, not ranked.

Top CPU score : Gautam- 11537 - i9 7960X @ 4.8 GHz
Top GPU score : Gautam - 4916 - GTX 1080 Ti
Lowest volume : Duality92 - 0.63L - ECS Liva Z Plus

 

[InfernoZeus]

Picture of actual PC

553211 - Duality92 - i3 8350K - GTX 1070 8GB - (1234*4321)/10 - #2 (Example only)

Requires photo in rules. Posts example without photo.

Nice idea though!

 

[jmarin]

I feel like budget should be incorporated somehow because some of us aren't willing/can't spend $600-$1000 on a CPU/GPU combo alone

 

[Duality92]

I feel like budget should be incorporated somehow because some of us aren't willing/can't spend $600-$1000 on a CPU/GPU combo alone

Performance is always about budget. Higher score = higher end parts = higher budget.

We see the parts used, so budget can be derived from MSRP of those parts

 

[GuilleAcoustic]

Power draw is also important. Saw many clusters moving to less hungry CPU (often Ultra-SPARK to Opterons), even if it meant performance losses.

 

[Duality92]

It is important, but sort of irrelevant for performance per liter. For example, I'm going to put my ECS Liva Z Plus engineering sample on the list tonight, it has a current unlocked i5 7300U (means it's always on boost at 3.5 GHz) with upgraded RAM. It draws 40W from the wall, but the performance reflects that.

It'd be 117 x 128 x 33 mm for 0.5L with 108 x 46 x 28 mm brick for 0.13L

400610 - Duality92 - i5 7300U Engineering Sample - Intel HD Graphics - (1051.6*240)/(0.5+0.13) - #6

https://valid.x86.fr/39sv36

Pictures : http://www.overclock.net/forum/18088-hardware-reviews-program/1631024-ocn-labs-ecs-liva-z-plus.html

Theoretically, the best you could have right now is a 7980XE on ASRock ITX with a 1080Ti on dual HDPlex Hi-Fi 400's, which would be possible in like 180*200*100 for under 4L.

 

[Curiosity]

Ill try to pull out and bench my Mini 3.24L tonight.
I'll edit this post once I do.
Dell da-2 is .87L
Total volume of 4.11L
??? - Curiosity - i5 3570K - ECGA 1060sc - ((CPUZMT*Heaven)/3.24L)) - Post #

https://valid.x86.fr/t7x45v
Having trouble getting myname to show up in the validx86 link. I guess I'm missing something obvious on how to do.
imgur.com/a/qDliCBQ

561911.3

 

[Duality92]

my official post done #6

 

[GuilleAcoustic]

Don't forget to put all external AC-DC bricks size is those PPL calculation ;-).

 

[InfernoZeus]

Don't forget to put all external AC-DC bricks size is those PPL calculation ;-).

And the controversy starts...

 

[Choidebu]

The real controversy is performance or footprint (e.g. power draw, cost, longevity)

The quest for performance for the sake of performance is, IMHO, silly.

I may be of minority here - I like to actually fully use my gear rather than letting these benchmark use them for my viewing pleasure.

 

[GuilleAcoustic]

This ain't controversy ... Because one could also run PSU + pump + res + huge radiator out of the chassis and claim to have a very small case.

Imagine a dual lga2011 like below, with ek-annihilators, Titans / Quadro / Tesla with single slot waterblock .... Run water-cooling gear and power brick outside of the chassis .... Insane PPL

So yes, external power brick must be added to chassis overall size.

 

[InfernoZeus]

This ain't controversy ... Because one could also run PSU + pump + res + huge radiator out of the chassis and claim to have a very small case.

I think there's an often unmentioned point in this discussion, which is why people are striving for smallness. For me, it's two fold, minimises desk space, and makes it more portable.

Moving the PSU out of the case benefits both of these (for my purposes). The PSU can sit on the floor (or under my desk), and when travelling, the case fits more easily into cases as the max dimensions are lower (with the PSU either in the same bag or another).

If it were practical to move the cooling out of the case, I wouldn't be against it either.

 

[GuilleAcoustic]

I think there's an often unmentioned point in this discussion, which is why people are striving for smallness. For me, it's two fold, minimises desk space, and makes it more portable.

Moving the PSU out of the case benefits both of these (for my purposes). The PSU can sit on the floor (or under my desk), and when travelling, the case fits more easily into cases as the max dimensions are lower (with the PSU either in the same bag or another).

If it were practical to move the cooling out of the case, I wouldn't be against it either.

I am a thin mini-ITX user and I am all for external PSU, for the exact same reason. But in all fairness, I think this should account for PPL calculation.

 

[Duality92]

I agree that we need to add the brick/bricks and any cooling, although I think that any cooling should've been considered from the start.

The real controversy is performance or footprint (e.g. power draw, cost, longevity)

The quest for performance for the sake of performance is, IMHO, silly.

I may be of minority here - I like to actually fully use my gear rather than letting these benchmark use them for my viewing pleasure.

Let me ask you a question then. Do you have a processor, GPU, PSU, storage of exactly the amount/size/performance you use your system for, or you have a bit of extra of everything, just in case? As enthusiast, overkill is basically the norm. I don't need an X299 PC, but I have one regardless.

 

[aquelito]

Nice idea @Duality92

To be thorough and to be able to compare each score, it seems to me that the PPL should also be balanced with temperatures and noise.

 

[GuilleAcoustic]

Nice idea @Duality92

To be thorough and to be able to compare each score, it seems to me that the PPL should also be balanced with temperatures and noise.

Maybe build a main indice that is PPL and then have sub-indices like PPL/W, PPL/C, PPL/DB, PPL/$

 

[Duality92]

But it doesn't work that way, we can be five with 8600K's @ 4.5 GHz but all of them will operate at different voltages and output different heat, requiring different fan speeds with the same cooler that we bought for different prices. I'd like to keep this thread simple, but if users would want to add more data I'm inviting someone else to manage the sub-indices that will have an overwhelming amount of data. If I put my PC outside for benchmarks it'd make me score higher for PPL/W, PPL/C and PPL/dB, that doesn't really make sense of what we're trying to achieve. My goal is to have a simple, easily manageable, not needing an overwhelming amount of data from users to participate, forum competition.

Alternatively, processors that output more heat at specific voltages, will be better at overclocking under good cooling because of the way processors work. With this in mind, a technically better potential chip will score lower, which is kind of the opposite of what we're trying to achieve.

 

[aquelito]

I get your point. There are indeed hundreds of possible data combinations.

However, to me, the finality of the PPL is to ultimately let you compare the viability of each design.

If your build can run pretty high scores but crashes after 10 min of benchmarks due to the heat or a underpowered PSU, the PPL is pretty pointless.

 

[GuilleAcoustic]

I get your point, but in a daily use point of view SFF users aren't running benchmarks 24/7. Metrics that do not represent real usecases are rather pointless.

I think it would be useful to know wether those 20% PPL improvement are worth the 60% price increase.

Wattage or heat is also important for stability. Let's say you want to build rendering nodes that will be running at full tilt for days .... Less heat means less risk of failure due to overheating.

A user can be after the best PPL within a budget, within a power draw (running from batteries), within a thermal envelope (hot weather, stability for long running computations)

This thread is a great idea, but I think side informations doesn't add much effort (even if you only compile PPL in the end). SFF members will have them at hand, should they need them.