Post your Cinebench R23 Score

[Deleted member 74752]

The 5700G is supposed to boost to 4.65 on stock settings, meanwhile your default ST run shows a lower result than with a manual overclock. I was initially surprised to see that my oc'd 3300X does better in ST than this Zen 3 APU. Why doesn't the 5700G reach its max clock with lightly threaded loads? Am I missing something here?

It will boost that high, but not on all cores...that was the reason for manual override. I must say I would not mind this chip in my gamer over the 3700X from what I have seen thus far. I did not try prioritizing cores in task manager...perhaps that would have been a different outcome.

 

[Ivan Bliminse]

http://imgur.com/TAyPypf

 

[harm9963]

Decided to run fans at 100 percent , for benchmarks its ok , not for everyday use.
Did 10min test as well, fans 100 percent for fun.

 

[Epic2360]

i9 12900K
NH-D15 cooler

 

[md2003]

12600k @5.4/5.2GHz passes 30' c23, 1h prime 95 small ffts, cooled by ac 420mm aio.

 

[AlwaysHope]

What kinda GFlops do these put out with Linpack Xtreme?

Just curious

Agree, that would be interesting.

 

[md2003]

@freeagent @AlwaysHope overview from system i am testing atm, GFlops are also shown.

 

[AlwaysHope]

@freeagent @AlwaysHope overview from system i am testing atm, GFlops are also shown.

Getting a little off topic for CB but...
Good job however we have to run the benchmark test not the stress test with that app to get more accurate figure of GFLOPS performance, that is according to the author of the app.
Probably best to post those scores in the LinpackXtreme thread.

 

[tabascosauz]

@rickss69 @Stormbringer not sure what's going on with these 5700G scores but the ST is quite low even at stock. I don't put much cooling on mine since it's in a tiny case but there's no reason to be dipping below a stock Zen 2 ST (~1300).

Mine is running some mild CO so ~4.78GHz ST, but even stock should not be below 1400 minimum to be performing properly.

Probably will be running scores on this 5700G again with a much larger C14S in the next few days, we'll see how things change.

 

[Deleted member 74752]

@rickss69 @Stormbringer not sure what's going on with these 5700G scores but the ST is quite low even at stock. I don't put much cooling on mine since it's in a tiny case but there's no reason to be dipping below a stock Zen 2 ST (~1300).

Mine is running some mild CO so ~4.78GHz ST, but even stock should not be below 1400 minimum to be performing properly.

Probably will be running scores on this 5700G again with a much larger C14S in the next few days, we'll see how things change.

View attachment 234799

At that frequency why is your multi-score not higher?

 

[tabascosauz]

At that frequency why is your multi-score not higher?

Benchmate records telemetry for that specific run. I made the screenshot after the ST test only, hence 61C and 4.8GHz and 44W. I can't remember but MT was probably around 4.3GHz all core??

5700G is noticeably more thermally dense than 5600G so stock tends to be more ST biased than MT. Which is why I was curious as to why your ST score was quite low compared to the good MT scores.

 

[Deleted member 74752]

Benchmate records telemetry for that specific run. I made the screenshot after the ST test only, hence 61C and 4.8GHz and 44W. I can't remember but MT was probably around 4.3GHz all core??

5700G is noticeably more thermally dense than 5600G so stock tends to be more ST biased than MT. Which is why I was curious as to why your ST score was quite low compared to the good MT scores.

Something I did not try, but can you test yours by going into Task Manager and prioritizing a particular core?

 

[tabascosauz]

Something I did not try, but can you test yours by going into Task Manager and prioritizing a particular core?

Ah...about that, Core Affinity has never worked for me in CB. It either doesn't do anything or crashes CB every time, across 5 different Ryzens. It shouldn't make too much of a difference as Cinebench is one of the better programs that should consistently identify your best cores. On my 3700X it was Core 5 and 7, 4650G was Core 0 and 5 I think, so they were really bad but CB had no problem. 5900X, 5600G and 5700G are pretty lucky and ideal, I think it's Core 0 and 1 across all 3 CPUs.

Boost algorithm+scheduler usually will juggle load every few seconds between your #1 and #2 rated cores in CB R20 and R23 ST test, if your firmware is all up to date and nothing is awry.

At stock settings, how do your clocks look on HWInfo in the 2 loaded cores? Your manual all-core OC results look spot on, so something must be wonky about the firmware.

 

[Deleted member 74752]

Ah...about that, Core Affinity has never worked for me in CB. It either doesn't do anything or crashes CB every time, across 5 different Ryzens. It shouldn't make too much of a difference as Cinebench is one of the better programs that should consistently identify your best cores. On my 3700X it was Core 5 and 7, 4650G was Core 0 and 5 I think, so they were really bad but CB had no problem. 5900X, 5600G and 5700G are pretty lucky and ideal, I think it's Core 0 and 1 across all 3 CPUs.

Boost algorithm+scheduler usually will juggle load every few seconds between your #1 and #2 rated cores in CB R20 and R23 ST test, if your firmware is all up to date and nothing is awry.

At stock settings, how do your clocks look on HWInfo in the 2 loaded cores? Your manual all-core OC results look spot on, so something must be wonky about the firmware.

I had Precision Boost disabled for my runs and manually overclocked. My only "tweak", if you can call it that was to go into Task Manager and run priority on "high" for Cinebench. I found that any app or process running in the background during the benchmark just kills the score. My guess is yours would be higher without that "Benchmate" attached.

I can confirm that affinity selection has no effect...just tested it.

 

[AlwaysHope]

Pleased with below 70C peak & utilizing MSI's "cpu lite load" option in bios (custom config for auto vcore with my particular silicon quality) , also the fact that 1.5v for DIMMs, 1.46v for VCCIO2 & 1.36v for SA - all points of additional heat within the system. Ambients of 24C.
Default CPU speeds.

 

[Deleted member 74752]

Pleased with below 70C peak & utilizing MSI's "cpu lite load" option in bios (custom config for auto vcore with my particular silicon quality) , also the fact that 1.5v for DIMMs, 1.46v for VCCIO2 & 1.36v for SA - all points of additional heat within the system. Ambients of 24C.
Default CPU speeds.
View attachment 234800

Now that you know your thermals are good, turn all that background junk off and run it for a better score.

 

[tabascosauz]

I had Precision Boost disabled for my runs and manually overclocked. My only "tweak", if you can call it that was to go into Task Manager and run priority on "high" for Cinebench. I found that any app or process running in the background during the benchmark just kills the score. My guess is yours would be higher without that "Benchmate" attached.

I can confirm that affinity selection has no effect...just tested it.

Completely agree. My 5700G is in my HTPC where I strip Windows down to the bare minimum necessary. Benchmate is okay - I guess maybe a point or two here and there lost. I find that the background programs really take their toll when running the ST test due to how long it is - the MT test is negligible.

 

[harm9963]

5950X has so much overhead room , its amazing , with a good AIO and ASUS DARK HERO ,only a few C higher than stock settings and 30 watts higher as well , score to temp ratio and watts , 5950X is a sleeper.

 

[AlwaysHope]

Now that you know your thermals are good, turn all that background junk off and run it for a better score.

Now your getting sophisticated, that's too much for little old me...
I don't like running any benchmark unless its truly indicative of what a system can do without software hacks. Take windows as it is imo.

 

[Deleted member 74752]

Completely agree. My 5700G is in my HTPC where I strip Windows down to the bare minimum necessary. Benchmate is okay - I guess maybe a point or two here and there lost. I find that the background programs really take their toll when running the ST test due to how long it is - the MT test is negligible.

I just tried msconfig and booted minimal...did not seem to have much effect. Currently using Ghost Spectre 10 Compact for the OS.

Now your getting sophisticated, that's too much for little old me...
I don't like running any benchmark unless its truly indicative of what a system can do without software hacks. Take windows as it is imo.

That's no software hack lol...just run the bench without all those other applications running in the background.

 

[AlwaysHope]

That's no software hack lol...just run the bench without all those other applications running in the background.

I always do with benching except thermal monitoring app. That is run windows vanilla.

 

[Deleted member 74752]

I would just like to get to the bottom of this ST low score for some of these 5700G's...makes me itch lol!

 

[Deleted member 74752]

Replaced the stock cpu cooler with a Noctua to eliminate that variable trying to get to the bottom of this 5700G single core performance...or lack thereof. I have tried all the tweaks I can come up with to no avail. I did manage to get the cpu to boost to 4850MHz in single core, but that is not the magic here. I've just about decided it is a limitation to some motherboards and or bios.

 

[tabascosauz]

Replaced the stock cpu cooler with a Noctua to eliminate that variable trying to get to the bottom of this 5700G single core performance...or lack thereof. I have tried all the tweaks I can come up with to no avail. I did manage to get the cpu to boost to 4850MHz in single core, but that is not the magic here. I've just about decided it is a limitation to some motherboards and or bios.

View attachment 235081

Seems like a good score if you're at stock? I'm guessing not stock if 4850, that's +200 PBO. The numbering for Asus BIOS is a bit different between B550 and X570 (I used to have the TUF B550M Wifi, but only used up to BIOS 2006 or so), but I think my Impact is on the equivalent of the 2403 BIOS for you (3601 for me). Could give it a try? Win 10 performance was a bit lower for me on any BIOS newer than 3601, AGESA every once in a while has these inexplicable performance regressions on newer BIOSes.

My 5700G is in the Impact already but I'm in the process of moving it into the Cerberus so I can mount my C14S. I'll run the bench and have some results for you soon.

 

[Deleted member 74752]

I was using the Noctua NH-D15 this evening @ 4800MHz...it had no problem keeping the temps in check. Both fans at max and I can barely hear them. The Wraith Prism would be screaming at me lol. I bought it for the gamer, but I hated how it jammed up the video card and covered the memory.