Post your Cinebench 2024 score

[iameatingjam]

View attachment 361165

This is the Windows install from my previous build with the 7900x, when I reinstall with a clean 24H2 I can take a look to see if there is a difference. I just need the time and desire to do it

Alright, but that signals to me that something is wrong. You may not even be getting the reported speed. Which would essential cut memory performance in half.

 

[AVATARAT]

Alright, but that signals to me that something is wrong. You may not even be getting the reported speed. Which would essential cut memory performance in half.

I checked some results with 9700x on hwbot, all there show the same.
Maybe CPU-Z needs an update?

 

[iameatingjam]

I checked some results with 9700x on hwbot, all there show the same.
Maybe CPU-Z needs an update?

idk maybe its a misreading but generally cpu_z is THE way to make to sure dual channel (or technically quad channel in ddr5s case) is working correctly.

@Dr. Dro knows a lot about memory. Maybe he can shine some light on this. In CPU_Z avarat is showing 2x32 bit channels. Does this not indicate that he's running in single channel mode (or dual channel when it should be quad technically?) Or is there a possibility its a misreading or is there some other explanation?

 

[mrcardio33x]

 

[AVATARAT]

9700x @ PBO
A little update:

 

[iameatingjam]

@AVATARAT

Looks like the memory channel thing might be a bug with certain versions of cpu-z

https://www.reddit.com/r/PcBuild/comments/16z0hgh
It is a year old, but still, maybe its not patched yet or maybe you're on older version.

 

[freeagent]

You are outperforming a Zen 3 12 core part, I would say that is pretty good

 

[mrcoffeetablebook]

Update to my previous post...multi coe has jumped up from 1320. I decided to ditch "Intel Default Settings" in bios

 

[iameatingjam]

Here's my single thread score for my 14700k on a new configuration that keeps vcore under 1.38, and of course all the other limits that come with the intel profile and turbo 3 turned off (no 5.6 boost) and reduced memory oc, not bad considering the average for this cpu is 131 and single core is when highest voltage is needed. I'm pretty happy with this Lost like 2%, could even just be run to run variance.

 

[ShrimpBrime]

14700K 8P 16T at 6,247Mhz. no e-cores. 285w peak.
New motherboard (yay!). 14th gen release bios and ME.
No degradation here. I don't know what all the hoopla is about. I constantly beat the shit out of this thing.
OH GOD ITS RUNNING 1.58v at only 800mhz. Who cares , all the cores are gated. Don't worry about that little guy right there.

 

[Burs]

I'm baffled as to why am I getting such an ultra-low score for single-core on my i7-14700K.
Sure I tweaked some BIOS settings, but I hadn't expected anything like this.
Quite on the contrary, I expected a better-than-average score.

Here's what I changed in my BIOS (everything else is on default):
Intel Default Profile (253W for both PL1/2, 307A)
CPU voltage offset -135mV
P-cores offset +2 (5.8Ghz)
E-cores offset +3 (4.6Ghz)

In MSI Afterburner I see Vcore around 1.24V, and the CPU is constantly clocked at 5.8Ghz.
Fans - I don't even hear them in this test as the CPU Package Power is only around 30W, and the CPU temp is below 50°C ?!

Rock-stable running literally anything I throw at it. Multi-core benchmarks are all completely fine, with max CPU temp in the low 90s.
It's just these single-core benchmarks in all Cinebench versions (R15/R23/2024)!
(in R15 I have a lower single-core score than i7 4th-gen, and in R23 lower than i7 7th-gen LoL! WTF?!)

Can anyone explain this? What am I doing wrong? I'm usually not such a noob, but for the life of me, I can't figure this one out.

 

[ShrimpBrime]

I'm baffled as to why am I getting such an ultra-low score for single-core on my i7-14700K.
Sure I tweaked some BIOS settings, but I hadn't expected anything like this.
Quite on the contrary, I expected a better-than-average score.

Here's what I changed in my BIOS (everything else is on default):
Intel Default Profile (253W for both PL1/2, 307A)
CPU voltage offset -135mV
P-cores offset +2 (5.8Ghz)
E-cores offset +3 (4.6Ghz)

In MSI Afterburner I see Vcore around 1.24V, and the CPU is constantly clocked at 5.8Ghz.
Fans - I don't even hear them in this test as the CPU Package Power is only around 30W, and the CPU temp is below 50°C ?!

Rock-stable running literally anything I throw at it. Multi-core benchmarks are all completely fine, with max CPU temp in the low 90s.
It's just these single-core benchmarks in all Cinebench versions (R15/R23/2024)!
(in R15 I have a lower single-core score than i7 4th-gen, and in R23 lower than i7 7th-gen LoL! WTF?!)

Can anyone explain this? What am I doing wrong? I'm usually not such a noob, but for the life of me, I can't figure this one out.

View attachment 362355

Maybe the single core is being run on an e-core.

Open CBR24 and then open task manager. Go to processes right click CBR24 and open to details. Set affinity to one of the p-cores, then set priority to high or real-time. If you use real-time, the rendering may just be blank until it finishes the benchmark, nothing to worry that's normal.

Good Luck.

 

[sophie]

14900ks

 

[Burs]

Maybe the single core is being run on an e-core.

Open CBR24 and then open task manager. Go to processes right click CBR24 and open to details. Set affinity to one of the p-cores, then set priority to high or real-time. If you use real-time, the rendering may just be blank until it finishes the benchmark, nothing to worry that's normal.

Good Luck.

Thanks for the reply.

Nice idea, I like your thinking. But unfortunately, that's not the case.
MSI AB shows the CPU is clocked @5.8, and my E-cores are @4.6, so the thread must be executing on one of the P-cores.

I've set the affinity to CPU0 anyway (assuming it's the P-core) and got an even lower score. Right when I thought it couldn't be worse LoL

But meanwhile, I think I found something, and it's not good. I was adding the offset voltage in +25 mV increments, and my single-core score was getting higher ever so slightly. When I reached an offset of 0 mV (no undervolt), my single-core score was more or less the same as the ones posted in this thread, but a multi-core test started throttling after 2 minutes only because I was getting into the 1.45V range with my Vcore).

I spent 2 days finding the right offset voltage (while -135 mV and 1.24 Vcore sure did look suspicious for 5.8/4.6Ghz), and I was only focused on stability and not paying attention to the benchmark scores. So now I need to find another way of lowering the voltage, and other options I see look much more complicated to me. Tho it's still funny why the scores go down when I'm applying my undervolt; maybe I'm just dumb. Seems it will be "back to school" for me over the weekend..

 

[Super Firm Tofu]

Thanks for the reply.

Nice idea, I like your thinking. But unfortunately, that's not the case.
MSI AB shows the CPU is clocked @5.8, and my E-cores are @4.6, so the thread must be executing on one of the P-cores.

I've set the affinity to CPU0 anyway (assuming it's the P-core) and got an even lower score. Right when I thought it couldn't be worse LoL

But meanwhile, I think I found something, and it's not good. I was adding the offset voltage in +25 mV increments, and my single-core score was getting higher ever so slightly. When I reached an offset of 0 mV (no undervolt), my single-core score was more or less the same as the ones posted in this thread, but a multi-core test started throttling after 2 minutes only because I was getting into the 1.45V range with my Vcore).

I spent 2 days finding the right offset voltage (while -135 mV and 1.24 Vcore sure did look suspicious for 5.8/4.6Ghz), and I was only focused on stability and not paying attention to the benchmark scores. So now I need to find another way of lowering the voltage, and other options I see look much more complicated to me. Tho it's still funny why the scores go down when I'm applying my undervolt; maybe I'm just dumb. Seems it will be "back to school" for me over the weekend..

A couple of things first. For accurate clock speeds, download HWiNFO, open to the Sensors Only Option, and watch 'Core Effective Clocks'. Afterburner isn't really as accurate. Second, download Benchmate as it includes a suite of all the current benchmark applications (including CB 2024). When done, it will give you a summary screen (by pressing F6) with accurate information after the run like @ShrimpBrime posted right above your first post.

Benchmate:

BenchMate

benchmate.org

HWiNFO:

Free Download HWiNFO Sofware | Installer & Portable for Windows, DOS

Start to analyze your hardware right now! HWiNFO has available as an Installer and Portable version for Windows (32/64-bit) and Portable version for DOS.

www.hwinfo.com

As for why you're single core score is so low, I'm thinking Undervolt Protection might be enabled in your BIOS. With that enabled, the higher your undervolt, the slower the CPU will run. Essentially, UV Protection slows the CPU down when it's not getting the voltage it thinks it needs to prevent it from crashing because of low voltage. It's probably why you are stable with such a large negative offset.

There's lots of way to adjust voltage, but I've always just set to adaptive mode with a negative offset. Each CPU will vary on how much offset is possible and still remain stable.

 

[ShrimpBrime]

A couple of things first. For accurate clock speeds, download HWiNFO, open to the Sensors Only Option, and watch 'Core Effective Clocks'. Afterburner isn't really as accurate. Second, download Benchmate as it includes a suite of all the current benchmark applications (including CB 2024). When done, it will give you a summary screen (by pressing F6) with accurate information after the run like @ShrimpBrime posted right above your first post.

Benchmate:

BenchMate

benchmate.org

HWiNFO:

Free Download HWiNFO Sofware | Installer & Portable for Windows, DOS

Start to analyze your hardware right now! HWiNFO has available as an Installer and Portable version for Windows (32/64-bit) and Portable version for DOS.

www.hwinfo.com

As for why you're single core score is so low, I'm thinking Undervolt Protection might be enabled in your BIOS. With that enabled, the higher your undervolt, the slower the CPU will run. Essentially, UV Protection slows the CPU down when it's not getting the voltage it thinks it needs to prevent it from crashing because of low voltage. It's probably why you are stable with such a large negative offset.

There's lots of way to adjust voltage, but I've always just set to adaptive mode with a negative offset. Each CPU will vary on how much offset is possible and still remain stable.

Yeah, I don't use Adaptive v-core at all. Ever. OK, maybe occasionally if I'm cooling restricted (air).

Think you're right. Uv protection probably enabled. If not, the cpu (ME) will probably act similarly with so much UV.

At stock, lowest all core loads I've witnessed with defaults is 1.288v. (That's 5.5ghz all p-core 4.3ghz e-cores)

 

[Burs]

A couple of things first. For accurate clock speeds, download HWiNFO, open to the Sensors Only Option, and watch 'Core Effective Clocks'. Afterburner isn't really as accurate. Second, download Benchmate as it includes a suite of all the current benchmark applications (including CB 2024). When done, it will give you a summary screen (by pressing F6) with accurate information after the run like @ShrimpBrime posted right above your first post.

Benchmate:

BenchMate

benchmate.org

HWiNFO:

Free Download HWiNFO Sofware | Installer & Portable for Windows, DOS

Start to analyze your hardware right now! HWiNFO has available as an Installer and Portable version for Windows (32/64-bit) and Portable version for DOS.

www.hwinfo.com

As for why you're single core score is so low, I'm thinking Undervolt Protection might be enabled in your BIOS. With that enabled, the higher your undervolt, the slower the CPU will run. Essentially, UV Protection slows the CPU down when it's not getting the voltage it thinks it needs to prevent it from crashing because of low voltage. It's probably why you are stable with such a large negative offset.

There's lots of way to adjust voltage, but I've always just set to adaptive mode with a negative offset. Each CPU will vary on how much offset is possible and still remain stable.

Thanks for replying. And for suggestions from both of you.

I have HWiNFO64 installed too, but I was watching the Active Clocks window (which shows the same clocks as Afterburner), instead of Core Effective Clocks. But you're right, effective clocks are something entirely different. They rarely go into the 4Ghz range, let alone in the high 5Ghz, while Afterburner shows my P-cores are always in the 5.7-5.8Ghz range and my E-cores @4.6Ghz. So that explains the loss in performance I guess.

As for the Undervolt Protection setting, it's actually *disabled* (by default). I checked that setting even before I did any undervolting as I know it's a crucial step to do. But even if that setting was enabled, that would still not explain why multi-core tests are all fine. Running all cores at such a large negative offset should be even harder than a single core, no?

I will try to undervolt using the adaptive mode, while checking the core effective clocks. And I will also check out BenchMate, tho I have many of these tools already installed.

Yeah, I don't use Adaptive v-core at all. Ever. OK, maybe occasionally if I'm cooling restricted (air).

Think you're right. Uv protection probably enabled. If not, the cpu (ME) will probably act similarly with so much UV.

At stock, lowest all core loads I've witnessed with defaults is 1.288v. (That's 5.5ghz all p-core 4.3ghz e-cores)

UVP is actually disabled, but I wonder what you meant with "cpu (ME)"? Intel Management Engine? I read that it indeed can conflict with undervolting.

While reading I also found out that some CPU microcode patch was disabling undervolting of the Intel CPUs in the past.. I just hope the recent microcode update (0x129) doesn't have something to do with my issue, because before I updated my BIOS last month, I could only set waaay lower undervolt. But could be something else too, as I've skipped a few BIOS versions before that one.

EDIT:

Actually, we're both right and wrong regarding the undervolt protection setting. Although the option is disabled, there's *another* related setting called "IA CEP" that has the same hate towards undervolting (there are some interesting articles about it on the net)

Now that I've disabled IA CEP, this is my new single-core score:

Me happy! 130%+ increase with only one option disabled Plus I had to dial down offset voltage by 20 mV.

 

[iameatingjam]

Thanks for replying. And for suggestions from both of you.

I have HWiNFO64 installed too, but I was watching the Active Clocks window (which shows the same clocks as Afterburner), instead of Core Effective Clocks. But you're right, effective clocks are something entirely different. They rarely go into the 4Ghz range, let alone in the high 5Ghz, while Afterburner shows my P-cores are always in the 5.7-5.8Ghz range and my E-cores @4.6Ghz. So that explains the loss in performance I guess.

As for the Undervolt Protection setting, it's actually *disabled* (by default). I checked that setting even before I did any undervolting as I know it's a crucial step to do. But even if that setting was enabled, that would still not explain why multi-core tests are all fine. Running all cores at such a large negative offset should be even harder than a single core, no?

I will try to undervolt using the adaptive mode, while checking the core effective clocks. And I will also check out BenchMate, tho I have many of these tools already installed.


UVP is actually disabled, but I wonder what you meant with "cpu (ME)"? Intel Management Engine? I read that it indeed can conflict with undervolting.

While reading I also found out that some CPU microcode patch was disabling undervolting of the Intel CPUs in the past.. I just hope the recent microcode update (0x129) doesn't have something to do with my issue, because before I updated my BIOS last month, I could only set waaay lower undervolt. But could be something else too, as I've skipped a few BIOS versions before that one.

EDIT:

Actually, we're both right and wrong regarding the undervolt protection setting. Although the option is disabled, there's *another* related setting called "IA CEP" that has the same hate towards undervolting (there are some interesting articles about it on the net)

Now that I've disabled IA CEP, this is my new single-core score:

View attachment 362447

Me happy! 130%+ increase with only one option disabled Plus I had to dial down offset voltage by 20 mV.

Did you have to disable the intel profile to disable CEP? We have to do that on gigabyte boards.... which kind of sucks because if you go back to gigabyte profile it changes everything back to how to it was before, which means a lot of tedius work putting all the settings back in safe(r) places. I think I'm gonna keep CEP on. Might be pointless but who knows, intel says keep it on so might as well I figure. There are ways to work around it.

 

[Burs]

Did you have to disable the intel profile to disable CEP? We have to do that on gigabyte boards.... which kind of sucks because if you go back to gigabyte profile it changes everything back to how to it was before, which means a lot of tedius work putting all the settings back in safe(r) places. I think I'm gonna keep CEP on. Might be pointless but who knows, intel says keep it on so might as well I figure. There are ways to work around it.

I'm still on Intel Default Settings. MSI board here. I was reading that some people don't even see the IA CEP setting. So I guess we're lucky we can disable it. Otherwise, I would have to find a different approach and retest everything which takes a ton of time.

 

[RootinTootinPootin]

 

[Daytrader]

Guys, i have a 13900K and using the latest intel fixed bios all on default for my asus motherboard, when i run cpu multicore or cpu single core tests using cinebench 2024, and monitor with hwinfo64, my temps never go over 80c, and my pcores never go over 5000, should they not be going to like 5800 if not being throttled ? or at least near, thx

 

[iameatingjam]

Guys, i have a 13900K and using the latest intel fixed bios all on default for my asus motherboard, when i run cpu multicore or cpu single core tests using cinebench 2024, and monitor with hwinfo64, my temps never go over 80c, and my pcores never go over 5000, should they not be going to like 5800 if not being throttled ? or at least near, thx

With all the new limits, power throttling does seem to happen more. Still that seems kinda low, With my 14700kf they usually dont go below 5.1 on those all core loads. How are the scores? Whats your power usage? As expected, or? How different is that temp from before? Were you limiting yourself to 253w before the update as well? Just trying to think if maybe something else is happening, like CEP triggering or something, but that would show in lower power usage, temperature and performance... not necessarily reported clocks.

Btw,, I've decided to keep all my pcores max frequency the same. These single core and dual core boosts can really increase voltage by a lot for not much gain. In my case, the performance difference is basically nothing. But I'm 5,5 vs 5.6. You're 5.4 vs 5.8 so maybe you'll see more loss. But still.. Might want to consider it.

 

[Daytrader]

With all the new limits, power throttling does seem to happen more. Still that seems kinda low, With my 14700kf they usually dont go below 5.1 on those all core loads. How are the scores? Whats your power usage? As expected, or? How different is that temp from before? Were you limiting yourself to 253w before the update as well? Just trying to think if maybe something else is happening, like CEP triggering or something, but that would show in lower power usage, temperature and performance... not necessarily reported clocks.

Thx for the reply, i am not great on stuff like this, my older/broken bios was also on default, i know the intel fix has changed stuff, i think this bios i am on now is 253w max for both of the options, i just thought that if me temps were ok i should get alot higher than like 4900/5000 on pcores, like i say i am not great on this stuff, just worried my cpu may be broken, only been used like a week or so, my scores were lower also than before, oh and temps before wers shooting to like 100c i am sure.

 

[iameatingjam]

Thx for the reply, i am not great on stuff like this, my older/broken bios was also on default, i know the intel fix has changed stuff, i think this bios i am on now is 253w max for both of the options, i just thought that if me temps were ok i should get alot higher than like 4900/5000 on pcores, like i say i am not great on this stuff, just worried my cpu may be broken, only been used like a week or so, my scores were lower also than before, oh and temps before wers shooting to like 100c i am sure.

Well its not unusual for boost clocks to fall below max in all core loads, when you're hammering every core there's only so much power to go around so that not necessarily a big issue and varies a lot from chip to chip ( the silicon lottery), you should still be able to reach max boost in most real world tasks. But how much lower are your scores? like 1-5%? I wouldn't worry. Or is it more than that?

 

[Daytrader]

was getting before bios update like 2200, now like 1800, 20% drop, seems alot, just watched a youtube video of someone using cinebench 2024, and his cores going to like 5500 5700 all the time and his temps 90 to 100, mine like 4800/5000, and temp never over 70/80, was as before was going in 90's to 100c, seems like something is holding the cpu back, surely the new intel bios would not be causing this ? thx