【Maxon】Cinebench R20 Benchmark Thread

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,432
136
@topmysteries5 , I can get a threadripper 1950x for $575 right now, and your CPU is $760 best price on ebay, so I would not consider it, seeing as the 16 core 1950x destroys the 18 core Xeons with modded bios.. I know I can get a few more mhz with modded bios, but I would rather spend my money on the next generation Ryzens, the current generation just beats Intel, and the next will destroy it.
And thans for the info "LFM bug, "lowest frequency mode" bug. ", but it was also only doing 50% or so utilization, in addition to the speed problem.
 

topmysteries5

Member
Jan 31, 2019
179
17
61
@topmysteries5 , I can get a threadripper 1950x for $575 right now, and your CPU is $760 best price on ebay, so I would not consider it, seeing as the 16 core 1950x destroys the 18 core Xeons with modded bios.. I know I can get a few more mhz with modded bios, but I would rather spend my money on the next generation Ryzens, the current generation just beats Intel, and the next will destroy it.
And thans for the info "LFM bug, "lowest frequency mode" bug. ", but it was also only doing 50% or so utilization, in addition to the speed problem.
Hi, i never said to upgrade your cpu, what i tried to show is real performance of these xeons. Anyway i got this at around half price which you mentioned.
 

SPBHM

Diamond Member
Sep 12, 2012
5,056
409
126
@topmysteries5 , I can get a threadripper 1950x for $575 right now, and your CPU is $760 best price on ebay, so I would not consider it, seeing as the 16 core 1950x destroys the 18 core Xeons with modded bios.. I know I can get a few more mhz with modded bios, but I would rather spend my money on the next generation Ryzens, the current generation just beats Intel, and the next will destroy it.
And thans for the info "LFM bug, "lowest frequency mode" bug. ", but it was also only doing 50% or so utilization, in addition to the speed problem.

if I force downclock my CPU via the windows power setting (like maximum 5%), it will also show lower % total utilization under full load, but that's I think based on 100% as full clock, so it's actually loading 100% of the cores but at the lower clock I think.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,432
136
Trying to use modded bios (which is sometimes problematic) and overclock my Xeons to equal STOCK second gen Ryzens, vs waiting a few months for the new Ryzens ? (I will say 3rd gen)... I will take the latter. Done with my Xeons for now. (until the 3000 series comes out)
 

Abwx

Lifer
Apr 2, 2011
10,847
3,296
136
Yeah sorry about that :D

Try again, using 256bit AVX2 ops Zen FPU utilisation goes up per thread as calculating 256bit ops takes twice the resources vs 128 bit ops. There's less FPU-resources left to have SMT gains. Intel's 256 bit vector units uses same amount of cpu resources whether using 128 or 256 bit ops but with 256 bit ops there's high possibility that same code uses less instructions with 256 bit ops -> more fpu resources available for SMT gains.

If there were that much AVX with an impact in MT then it would have an influence in ST as well, but to the contrary Intel lead wich was 5-6% in R15 shrinked to 2% or so.


Because if less instructions are required for a given troughput this save cycles to execute more instructions within a given numbers of cycles, be it in ST, so so much for the nice try, you demonstrated the contrary of what you intended at the first place...
 

AdamK47

Lifer
Oct 9, 1999
15,168
2,826
126
9900K @ 5050MHz two cores, 4950MHz four cores, and 4850MHz eight cores.

2fOpnGx.png
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
Broke the 4K mark with 5.2GHz, no AVX offset.

4061 cb
541 cb (single core)

Cinebench-R20-5200-MHz-SC.png


Takes almost 1.400V to achieve a stable 5.2GHz and nearly adds as much additional power usage as it took to go from 4.7GHz --> 5.1GHz, so I'm backing down to 5.1GHz for my 24/7 overclock.
 
  • Like
Reactions: lightmanek

SPBHM

Diamond Member
Sep 12, 2012
5,056
409
126
There s a summary at Computerbase.de using datas from their community.

https://www.computerbase.de/2019-03/cinebench-r20-community-benchmarks/

What is to be noticed is not only SMT scaling being lower than in R15 but also that the tables have been turned , with Intel suddenly scaling better, this has nothing to do with instructions but with specific optimisation of the code to suit better a uarch than the other, at least in this specific scene.

G5500 = 903
i3 6100 = 907

it looks like AVX is not a big deal since the G5500 lacks it, but it helped overcome 3.7 to 3.8GHz difference (and the extra 1MB l3 cache)
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
G5500 = 903
i3 6100 = 907

it looks like AVX is not a big deal since the G5500 lacks it, but it helped overcome 3.7 to 3.8GHz difference (and the extra 1MB l3 cache)
Something doesn't make sense here...
The g5500 only has a multi result but if you look at the difference between the i3-6100 and the g4560 the i3 is 9% faster in multi but then in single threaded the i3 is 40% faster...
 

topmysteries5

Member
Jan 31, 2019
179
17
61
You did see thats its identified as "LFM bug, "lowest frequency mode" bug. " and I tried my 2683v3 retail,, and the problem is not there.
I think this LFM bug can be fixed. When i was turbo modding my 2686v3, i did face this same bug on pc startup (cpu speed got stuck at X8, that is 800mhz).
After doing lots of experiments with efi file, i fixed it.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,432
136
I think this LFM bug can be fixed. When i was turbo modding my 2686v3, i did face this same bug on pc startup (cpu speed got stuck at X8, that is 800mhz).
After doing lots of experiments with efi file, i fixed it.
This is different. Just booting, it runs all cores 2.5 ghz. Until I run cinebench 20. The it goes to 1.2 ghz, and is stuck there no matter what until I reboot. But my retail 2683v3 does not have that problem, just the 2695v3 ES.
 

topmysteries5

Member
Jan 31, 2019
179
17
61
This is different. Just booting, it runs all cores 2.5 ghz. Until I run cinebench 20. The it goes to 1.2 ghz, and is stuck there no matter what until I reboot. But my retail 2683v3 does not have that problem, just the 2695v3 ES.
What speed you get on 2683v3 while running cinebench r15 and R20? use Hwinfo64/CPU-z to check you uncore and core speed while benchmarking.
E5 2683v3 runs all core x30 on normal load and x29 on AVX2 load. Cache/uncore must be 3ghz for best performance, it shouldn't drop during benchmarks

You can try same on ur 2695v3 ES also
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,432
136
What speed you get on 2683v3 while running cinebench r15 and R20? use Hwinfo64/CPU-z to check you uncore and core speed while benchmarking.
E5 2683v3 runs all core x30 on normal load and x29 on AVX2 load. Cache/uncore must be 3ghz for best performance, it shouldn't drop during benchmarks

You can try same on ur 2695v3 ES also
All of mine run 2.5 ghz, as I said. No modded BIOS.
 

topmysteries5

Member
Jan 31, 2019
179
17
61
All of mine run 2.5 ghz, as I said. No modded BIOS.
No modded BIOS means, under AVX2 load your cpu speed is dropping to 2.2GHz or less.
Without bios mod my 2686v3 drops to 2.4Ghz on normal load and 2.1Ghz on AVX2 load (cinebench R20).
 

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
2600X stock:

Single core: 514 Was probably 414. I ran it again and got 408 and 410 on another run.
SMT off: 2318
SMT on: 2986
 
Last edited:
  • Like
Reactions: IEC

naukkis

Senior member
Jun 5, 2002
701
568
136
If there were that much AVX with an impact in MT then it would have an influence in ST as well, but to the contrary Intel lead wich was 5-6% in R15 shrinked to 2% or so.


Because if less instructions are required for a given troughput this save cycles to execute more instructions within a given numbers of cycles, be it in ST, so so much for the nice try, you demonstrated the contrary of what you intended at the first place...

With 256 bit ops Intel hardware has double load/store bandwith to L1 cache vs 128 bit ops. Obviously L1 bandwith isn't problem for one thread as Zen can keep up but for SMT scaling extra L1 bandwith sure helps.
 

SPBHM

Diamond Member
Sep 12, 2012
5,056
409
126
Something doesn't make sense here...
The g5500 only has a multi result but if you look at the difference between the i3-6100 and the g4560 the i3 is 9% faster in multi but then in single threaded the i3 is 40% faster...

the single thread score for the G4560 looks wrong,
that would be a 3.2x scaling from a 2c/4t CPU.

the other CPUs with 2c/4t and fixed max clock on the list
200GE
301/758 (2.5x)

i3 3210m
231/534 (2.3x)

Core i3-2367M
101/250 (2.5x)

in any case, it certainly would be interesting to investigate more, but with the 4560,5400 and the 6100 it looks like AVX is not a big factor.