-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Intel Core i9-12900K allegedly scores 30K points in Cinebench R23
- Thread starter Dave2150
- Start date
Caporegime
Yeah my sample doesn't like going over 4.9 so its been slightly more optimized for multi threaded.
Thats still 3% higher than CPU-Z say it should score, what did their sample run at? 4.7Ghz - 4.8Ghz would be the clock speed for how they have it scoring. I did find it odd that they have the 5900X and 5950X scoring the same as the 5800X, they run 150 - 200Mhz higher clock speed ST out of the box. Even the 5800X is low at 4.7 - 4.8Ghz.
Soldato
- Joined
- 7 Dec 2015
- Posts
- 3,034
If the blue bar are about the same length as the pink bar as shown in the allegedly leaked screenshot, then does it mean that the 12900K and the 5950X have very similar multi-thread performance?
Possibly but i do not trust those CPU Z results. They have 5950X at 648 ST, mine gets 677 (4.85Ghz), the mT is even worse with them listing 11906 when mine scores 13385 (@4.65Ghz) and that's with rubbish memory running at 3200 CL16. This is why i will wait for real results from actual reviews
EDIT - Sorry i just seen they are only testing 24 cores of the 5950X so yeah they probably are very close but that would also tell me that the single threaded performance of Alderlake does not carry over to good multi threaded performance
Last edited:
Caporegime
Those Sandra Benches are a 5900X.
In a passing conversation a friend said Intel had paid CPU-Z "how to do benchmarks properly" his words, i didn't think to push him on that or where he got that info from.
But, how long has Zen 3 been out? for all that time CPU-Z never added any Zen 3 chips in the drop-down comparison, except very recently in the last patch or two, just before Alderlake is out.
Intel do have a past history of paying software vendors to gimp a competitors CPU or favour theirs, one of the most well known examples is Cinebench, but there are a lot more examples of it, they have been sued many times by all sorts of people for doing it.
If it turns out to be very good in Cinebench and CPU-Z but not much else, i would be very suspicious of it all.
In a passing conversation a friend said Intel had paid CPU-Z "how to do benchmarks properly" his words, i didn't think to push him on that or where he got that info from.
But, how long has Zen 3 been out? for all that time CPU-Z never added any Zen 3 chips in the drop-down comparison, except very recently in the last patch or two, just before Alderlake is out.
Intel do have a past history of paying software vendors to gimp a competitors CPU or favour theirs, one of the most well known examples is Cinebench, but there are a lot more examples of it, they have been sued many times by all sorts of people for doing it.
If it turns out to be very good in Cinebench and CPU-Z but not much else, i would be very suspicious of it all.
This would not surprise me based on Intels past record. I still do not think that Intel CPU Z shows good scaling from Single Thread to multi threaded use though. 20% higher than 5950X on single thread but the same (ish, score blurred out) when both at 24 cores
The CPU-Z bars for multi-threaded performance are FAKE. They always show full bar above a certain score line. Which is just stupid by its author. Alongside other things...
Caporegime
24 Threads, 12C-24T
No you're right it doesn't, the 5900X still wins most of it and where it doesn't its mostly a draw.
And the thing about so called "leaks" its Cinebench and CPU-Z that's being pushed out all over the place over and over again, it was the same with Rocket Lake and they don't do quite so well in most things outside CPU-Z and Cinebench, another thing that's similar to Rocket Lake is preorders, you can buy Alder Lake a week before the NDA for reviews lift but you wont actually get the CPU you bough until the review NDA lifts.
It seems to me they have the same confidence in Alder Lake as they had in Rocket Lake.
Thats still 3% higher than CPU-Z say it should score, what did their sample run at? 4.7Ghz - 4.8Ghz would be the clock speed for how they have it scoring. I did find it odd that they have the 5900X and 5950X scoring the same as the 5800X, they run 150 - 200Mhz higher clock speed ST out of the box. Even the 5800X is low at 4.7 - 4.8Ghz.
My 10900K runs the SC CPU-Z at 635 ish, so those scores need some work.
I don’t think so. The 5950X is super effective with SMT and it’s a full fat 32t chip. Intel will probably have a edge with upto 4 cores from each cluster loaded up. Essentially Intel are bringing out 11th gen chips with layers of HT.
Any 8 of Intel’s core vs AMD’s will be slower. The low power cores will probably pull less power than a 11900, but be also be significantly slower. The big cores will probably be similar in performance and fast across 4 cores.
As per alder lake seeming to fall behind in some multithread benches.
We should remember that much of Alder Lake's multi thread performance is from Thread Director and the Windows Scheduler.
And so it can be that certain software is unable to take advantage of the 8 little cores until they receive software updates, alternatively thread director may need updates through new firmware or bios to better work with certain software.
We should remember that much of Alder Lake's multi thread performance is from Thread Director and the Windows Scheduler.
And so it can be that certain software is unable to take advantage of the 8 little cores until they receive software updates, alternatively thread director may need updates through new firmware or bios to better work with certain software.
Thread director will likely need a lot of polish to work well. Middleware will likely set the speed limit for the most part.
Plus you have the hyper threading and ring bus exploits to consider, with Intel’s HT being dead for a number of years and being very hit and mis.
Impressive.
Looking forward to gaming benchmarks, be interesting to see how DDR5 affects things.
Soldato
- Joined
- 31 May 2009
- Posts
- 21,257
It is a little odd, as it seems to lose in every single one of the actual benchmarks, except the single core thing where it seems to be 5% faster. This will be interesting to see when it actually arrives and we see how it is setup.
Windows 11, PCIE 5.0, DDR5, 12th gen. It’s going to be very interesting. Unless you are on an old platform and need the upgrade, wait until 13gen next year
It's future proofed for sure, when alder lake launches there actually won't be any pcie5 graphics cards or ssds on the market, not for several months in fact - this begs the question how Intel tested its pcie5 support, chances are it's buggy and will need an updated bios once products release
PCIe 5 SSDs won't be reasonable for at least a year if not 2-3 years more.
Don't rush.
PCIe Gen 5.0 SSDs With Phison's Next-Gen E26 Controller To Start Shipping in 2022, Up To 16 GB/s Speeds (wccftech.com)
Raptor lake will launch next year, and will up the core count, frequency and IPC over Alderlake - that may be the one to get for most, to avoid the early adopter price premiems of DDR5 etc. Plus by the launch of Raptor Lake (rumoured Q2-Q2 22) windows 11's hardware scheduler should have had any bugs fixed and general performance improvements.
Raptor lake should be competing against Zen3 refresh (assuming Zen3 refresh isn't further delayed) - should be an interesting duel
DDR5 memory is looking like good for overclocking
in this leak, it's a 12900k on a z690 gigabyte board. The memory is a 4800 CL38 kit but the person has overclocked it to 8000 CL50.
what we do know from other leaks is that while DDR5 overclocks well it requires high voltages, we know even 10,000mhz is possible on DDR5 but you need 2.0v to do that, the question is how hot does it get with those voltages
https://wccftech.com/intel-core-i9-...0-aorus-tachyon-motherboard-ddr5-8000-memory/
in this leak, it's a 12900k on a z690 gigabyte board. The memory is a 4800 CL38 kit but the person has overclocked it to 8000 CL50.
what we do know from other leaks is that while DDR5 overclocks well it requires high voltages, we know even 10,000mhz is possible on DDR5 but you need 2.0v to do that, the question is how hot does it get with those voltages
https://wccftech.com/intel-core-i9-...0-aorus-tachyon-motherboard-ddr5-8000-memory/
Soldato
- Joined
- 31 May 2009
- Posts
- 21,257
CL50? ouch