Discussion Cinebench 2024 Released

[Markfw]

Supposedly there is INi file that can turn off some SIMD stuff:

https://twitter.com/x/status/1699307923929784787

So playing with it could reveal "speed up" factor.

I tried this, and it slowed it down. That is enabling things like avx and sse1, etc.... it slowed down about 3%. Maybe someone else who knows better can play with enabling something, but I am done.

 

[Mopetar]

Maybe in the next CB, they decide to include a heterogeneous test using all available CPU/GPU cores. The decision to use a highly parallel algo to maximize GPU parallelism has the unfortunate side effect of making all existing CPU cores look really wimpy.

Wouldn't it basically just be adding the two scores together? GPUs are just better at the workload so even lumping in the CPU cores wouldn't move the needle much unless you've got a 128 core server chip paired with a 1030 or something like that.

I guess it would be good for thermals testing or seeing how an APU performs under near 100% load on both the CPU and GPU.

 

[Saylick]

View attachment 85471

The graphics test didn't get the fans to ramp up either. The CPU use was only ~5% across the system during the run so there was probably plenty of thermal headroom.

Pretty dang good showing for Apple cores here.

 

[Mopetar]

Yeah they definitely punch above their weight class in CPU. But the dedicated video encoder hardware is really impressive.

The last Intel Mac I had could render video reasonably well and could spit out decent quality encodings at a littler under one minute per minute of video. You could tell it was hard at work because it didn't take even a minute for the fans to be going full throttle.

The MBP I have now will do even better quality in a few minutes and you the lack of fans still makes me think it hasn't actually started the encoding yet by the time it's already done.

I remember using some of the old POWER Macs where you'd just set it to render/encode overnight because it would take that long, but even that was considered fast for 1080p video.

 

[Iron Woode]

I decided to try my R5 3600 system:

 

[Orfosaurio]

Maybe in the next CB, they decide to include a heterogeneous test using all available CPU/GPU cores. The decision to use a highly parallel algo to maximize GPU parallelism has the unfortunate side effect of making all existing CPU cores look really wimpy.

Isn't Cinebench a benchmark for the capabilities of a PC for running Maxon software instead of a generalistic benchmark despite how they sell it?

 

[Markfw]

Isn't Cinebench a benchmark for the capabilities of a PC for running Maxon software instead of a generalistic benchmark despite how they sell it?

Its generally accepted that its a good multicore benchmark

 

[Abwx]

Its generally accepted that its a good multicore benchmark

For FP eventually but it s useless fon INT, in this latter case 7 ZIP is infinitely more relevant, previously AT used it as a benchmark for servers ST and MT expected perfs.

 

[Thunder 57]

Seems nobody's bothered to bench an AMD GPU so I'll add the pathetic score of my RX 5700: (3246)




I was also surprised by how it didn't spin up the fans at all. It seemed to be under full load though according to GPU-z.

 

[NikosD]

Have you seen any report or documentation of what kind of FP precision Cinebench 2024 is using for rendering ?

Is it FP32 or FP64 ?

Judging by the GPU results it should be using FP32 for GPUs, but is there any chance that is using FP64 for CPUs ?

It should be the same FP precision across all devices and platforms (SSEx, AVX2, Mac, Arm, GPUs etc)

 

[Hulk]

I miss the way R23 would show during the render a box indicating the progress of each thread. For my 13600K it shows two rendering boxes? What is the methodology behind the number of rendering boxes in R24?

 

[igor_kavinski]

For my 13600K it shows two rendering boxes?

It was the same when I ran it on a 48 thread Xeon server.

 

[Hulk]

It was the same when I ran it on a 48 thread Xeon server.

I like the old behavior. With my 13900K it was fun to watch all 32 threads render like a swarm of bees!

 

[Hail The Brain Slug]

Stable daily setup

 

[igor_kavinski]

I like the old behavior. With my 13900K it was fun to watch all 32 threads render like a swarm of bees!

True. After seeing 48 threads dance on the screen, I had a really strong craving for a 64 thread Epyc. I haven't tried the GPU test yet so maybe it is different and may show that previous swarm behavior?

 

[Hail The Brain Slug]

True. After seeing 48 threads dance on the screen, I had a really strong craving for a 64 thread Epyc. I haven't tried the GPU test yet so maybe it is different and may show that previous swarm behavior?

Nope, still individual squares.

 

[igor_kavinski]

Not a single 13900K(S) score. Where are all those users?

 

[igor_kavinski]

Nope, still individual squares.

Is it possible to disable AVX-512 in the BIOS? Was curious if lack of AVX-512 has a discernible impact on Zen 4 score.

 

[Thunder 57]

Is it possible to disable AVX-512 in the BIOS? Was curious if lack of AVX-512 has a discernible impact on Zen 4 score.

Pretty sure Cinebench doesn't use AVX-512. At least, that used to be the case.

 

[Hail The Brain Slug]

Is it possible to disable AVX-512 in the BIOS? Was curious if lack of AVX-512 has a discernible impact on Zen 4 score.

Last time I looked I was not able to find any toggle to disable AVX support in the UEFI.

 

[Seba]

Ryzen 5 3600 (with 32GB DDR4 3200)
CPU Multi Core: 538
CPU Single Core: 74

Radeon RX 7600 8GB
GPU: 3842

 

[Det0x]

Last time I looked I was not able to find any toggle to disable AVX support in the UEFI.

AGESA 1.0.0.8


And my maxed 7800x3d score:
MT = 1263 points
ST = 122 points


Maxed 7950x3d:
MT = 2396 points
ST = ~131 if i remember correctly

 

[Hail The Brain Slug]

AGESA 1.0.0.8
View attachment 87092

And my maxed 7800x3d score:
MT = 1263 points
ST = 122 points
View attachment 87093

Maxed 7950x3d:
MT = 2396 points
ST = ~131 if i remember correctly
View attachment 87095

Ah I'm not on 1.0.0.8 yet, on my board on whatever older agesa I was on when i checked, it didn't have any toggle. I'll look again after I update.

 

[AdamK47]

My 7950X3D with the HF CCD disabled. DDR5 @ 6200 and FCLK @ 2100. Same settings I've been using daily for many months to game.

Edit - Clean Run:

 

[Muadib]

I feel dumb, but I have to ask. How do I get this to run? I downloaded, unzipped, and double clicked on Cinebench.exe, to no avail. What am I missing? This is on the 5900x rig in my sig.