News 7-Zip Benchmark: Intel Core i9-13900K 60% Faster Compared To 12900K

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

salgado18

Distinguished
Feb 12, 2007
928
373
19,370
12900k 8core=1511:8cores=188=6% less performance than single thread.
(which can be compensated for with a bit more power which we can see form the O/C numbers)
New zen 4 (might not be accurate) 1329:8=166=30% less performance.
(And that's just 8 cores, not the full 16 cores difference)
https://cpu.userbenchmark.com/Compa...ed-Marketing-Devices-7600X/m1685583vsm1898605
1epa5P2.jpg
Wait, did you compare the top of the line 16-core i9 against the 6-core midrange Ryzen 5? It may be the number we have, it may be someone else's comparison, but you did write both numbers saying it's a 30% difference even though they are completely different.

Here's the fair comparison: the supposed Ryzen 7600X versus the i5-12600 (non-K, with 6 p-cores and no e-cores). This puts 6 cores against 6 cores. The Ryzen gets 1329 points, the i5 gets 1063 points.

https://cpu.userbenchmark.com/Compa...00X-vs-Intel-Core-i5-12600/m1898605vsm1751978

Or take the top i5 versus the Ryzen: 1244 points against 1329, now using two e-cores to help.

https://cpu.userbenchmark.com/Compa...X-vs-Intel-Core-i5-12600KF/m1898605vsm1700734

Always use apples-to-apples when possible, and in this case, it is possible.
 
  • Like
Reactions: bit_user
Wait, did you compare the top of the line 16-core i9 against the 6-core midrange Ryzen 5? It may be the number we have, it may be someone else's comparison, but you did write both numbers saying it's a 30% difference even though they are completely different.

Here's the fair comparison: the supposed Ryzen 7600X versus the i5-12600 (non-K, with 6 p-cores and no e-cores). This puts 6 cores against 6 cores. The Ryzen gets 1329 points, the i5 gets 1063 points.

https://cpu.userbenchmark.com/Compa...00X-vs-Intel-Core-i5-12600/m1898605vsm1751978

Or take the top i5 versus the Ryzen: 1244 points against 1329, now using two e-cores to help.

https://cpu.userbenchmark.com/Compa...X-vs-Intel-Core-i5-12600KF/m1898605vsm1700734

Always use apples-to-apples when possible, and in this case, it is possible.
Right I thought it was the 16core zen4, disregard that part.

But I was talking about how much the single threaded score changes the more cores are loaded, not about the total throughput.
 
  • Like
Reactions: salgado18
No matter how much against AMD they are their numbers have never been put in question, just the opinion they draw from it and how they choose to talk about it.
If being openly against a company disregards their numbers we should also disregard anything GamersNexus or hardware unboxed or (insert name here) puts out.
Your Internet must be different to mine, LOL.


They don't "choose" to talk about it; they shut down all conversations about why they're biased. They just hate AMD for some reason and won't straighten things up. This is no secret in the tech world and only people with the same biases defend UB.

Again, please do reconsider.

Regards.
 
Your Internet must be different to mine, LOL.


They don't "choose" to talk about it; they shut down all conversations about why they're biased. They just hate AMD for some reason and won't straighten things up. This is no secret in the tech world and only people with the same biases defend UB.

Again, please do reconsider.

Regards.
Oh well if gizmosphere...a site nobody ever heard of... says it it must be true.

So let us see what they claim:
So they are saying that UB favors single core and gives negative score to octacores and as proof of that they show a comparison of the 3900x getting almost twice the score of an unlocked i3.
Is the 3900x going to feel twice as fast when opening webpages or playing games?
One of the most glaring inaccuracies in UserBenchmark’s system is its favoring of single-core and quad-core processors, with octa-core processors being given negative assessments despite the fact that the latter objectively pumps out more power and speed. It took the Ryzen 5’s launch –which many users used as a new benchmark of sorts –to figure out the inaccuracies presented in UserBenchmark’s results.
Userbenchmark-Single-thread-performance-vs-multi-thread-performance.png


They just hate AMD for some reason and won't straighten things up.
They don't hate AMD, they also give moarcoars unlocked intel CPUs a worse score than lower cored locked models.
They are called USER benchmark because they are targeting USERS and not workstations and USERS are going to use less cores and after a number of them no amount of more cores is going to make any difference to them.

PbVH4qs.jpg
 
Oh well if gizmosphere...a site nobody ever heard of... says it it must be true.

So let us see what they claim:
So they are saying that UB favors single core and gives negative score to octacores and as proof of that they show a comparison of the 3900x getting almost twice the score of an unlocked i3.
Is the 3900x going to feel twice as fast when opening webpages or playing games?

Userbenchmark-Single-thread-performance-vs-multi-thread-performance.png



They don't hate AMD, they also give moarcoars unlocked intel CPUs a worse score than lower cored locked models.
They are called USER benchmark because they are targeting USERS and not workstations and USERS are going to use less cores and after a number of them no amount of more cores is going to make any difference to them.

PbVH4qs.jpg
Suit yourself then.

Regards.
 
So how does that change the fact that UB also values the i3 higher than the i9?
Actually they make the exact same point as I do.
This has riled AMD fans, but the issue actually impacts the rankings for all processors, as evidenced by the test result above. Due to the $120 quad-core Intel Core i3-8100's higher boost frequency, it beats Intel's $2,000 18-core 36-thread Core i9-9980XE in the rankings. It doesn't take a degree in computer science to see the obvious disconnect here.
The only thing in question is if simple users will see any benefit from more than 8 cores or not.
If you believe that more then 8 cores is important then all you have to do on the UB webpage is to slightly curl up your finger that is over the scroll wheel until you reach the 64-core score, you can disregard anything the site says and base your decision on that score alone.
 
  • Like
Reactions: KyaraM
No, the numbers in screenshot you're showing are based on user votes, they has nothing to do with the actual benchmark scores. The i9 outscores the i3 in UB benchmarks (as you'd expect).
Because the gizmo site also only showed that.

Again scroll a little lower and you will see the "value" score, the i3 is 21% higher valued, there are only those two scores on the page, user score and value, everything else is a benchmark number.
And while the 12900k wins in every single benchmark the i3 still wins out in the value score because of the huge price difference and the small difference in gaming, especially if you can't spring for a $2000 GPU.

If the argument here is that UB influences the noob PC users then the user score is all such a person would see and as such the only thing that could influence them.
 
  • Like
Reactions: KyaraM

Ogotai

Reputable
Feb 2, 2021
323
219
5,060
There are no worse hypocrites on the planet than AMD fboys.
the same can be said about the intel fanboys, pc warrior. after all they were the ones that would tout how well intel was doing in benchmarks ( like cinebench for example ) when intel was winning them, then down played those SAME benchmarks when intel started to loose on those very same ones. heck even intel them selves were downplaying benchmarks as a whole a a year or 2 ago when they were losing. just recently, the intel fanboys kept saying how important single core performance is, then one who them selves kept posting how single core performance was said this : " Who cares how well a single core does in an app that you will never run on only a single core? " so its fair to say both sides have their hypocrites, but it would seem now, intel has the most. i cant wait to see as well, what the intel side, who just last year i think it was, kept talking about how important AVX 512 is when intel had it and was winning benchmarks cause of it, but now that amd is also adding it with zen 4, and intel seems to have dropped it, how unimportant it will be.

I wait to see what they will say about AVX512 and its relevance now that the roles are reversed. I am willing to bet that they will all be cheering for the few benchmarks that use it and AMD wins
you mean, JUST LIKE the intel side was doing ?
 
There's no way a doubling of the eCores from 8 to 16, or 33% increase in thread count from 24 to 32, is the primary reason for a 60-70% increase in performance for this benchmark.
Supposedly, not only E-core counts will be increasing, but also clock rates and the amount of cache. If some recent leaks are to be believed, all-core boost clocks on the P-cores might be rising by more than 10%, and by more than 15% on the E-cores compared to the 12900K. Add to that a substantial increase in cache, supposedly more than double, and those numbers might not be too surprising for a heavily-multithreaded, cache-heavy workload like that. I would expect far lower gains in most other workloads, but multithreaded performance might see a decent boost.

So now that Intel takes the lead in yet another AMD stronghold, it of course gets downplayed and becomes unimportant and not quite “real world” for the AMD fbs.
A 60% gain in 7-Zip decompression wouldn't even necessarily be taking the lead though. At least according to this review, the Ryzen 5950X already performed over 50% faster than the 12900K at that test, and the real competition will be from Zen4, not hardware released two years prior...
https://www.techpowerup.com/review/intel-core-i9-12900k-alder-lake-12th-gen/13.html

No matter how much against AMD they are their numbers have never been put in question, just the opinion they draw from it and how they choose to talk about it.
Honestly, it's hard to take them seriously. I like the idea of the site, that you can easily get a rough idea of how hardware performs from completely different generations that might not be directly compared in reviews, and I also like the convenience of their web-based benchmark to determine if a particular piece of hardware might be underperforming compared to other systems with the same components. But they are completely unprofessional and appear to be blatant fanboys who adjust their weighting to favor a particular company, which makes them really difficult to recommend. I wouldn't be surprised at all if they hold a monetary interest in certain companies and do that as a means to help influence stock prices or something, because it just seems way too blatant.

Or maybe they just do that to distract from the fact that their benchmarks have flaws, and don't always meaningfully relate to different architectures. When their comparisons don't accurately reflect real-world performance differences between hardware, their response is to repeatedly claim that all the review sites testing real software are wrong, and double down on their rhetoric that they are the only place to get accurate performance numbers from, despite their tests being completely synthetic.
 
" Who cares how well a single core does in an app that you will never run on only a single core? "
No, because single threaded IS important it IS important to keep a high single threaded score even when all cores are doing work and not lose 25-30% of the single threaded score if other cores do some work as well.
You are not buying a 16core CPU to shut down 15 of the cores and only use one.

You will never run Cinebench or whatever multithread app just on a single core if you have a choice about it, but if a CPU cheats and gets 25-30% higher score in cinebench on a single core compared to all cores running then why not point it out that it's cheating?
 
  • Like
Reactions: KyaraM

bit_user

Polypheme
Ambassador
Until Intel offers 16 performance cores, AMD is leader.
Intel's thinking is half-az. Those efficient cores would mean something if they could match the performance core count of AMD. Until they do, they will be recognized as crap cores.
Running a thread on an E-core is faster than having it share a P-core with another thread. Having 16 of them should actually perform better than simply having 8 more P-cores, if you can use them all.
 

bit_user

Polypheme
Ambassador
1epa5P2.jpg



Also here is a benchmark with the 12900k locked to 125W as well as "stock" at 240W so it doesn't matter how many hours they would run they would always use that amount of power and get that amount of performance, as you can see the performance difference is 0% at single and about 10% - 20% in multithreaded while the power increase is 100%
Why do you say 12900K, but you selected 12900KF? I get that the difference is only a few %, but still...
 

bit_user

Polypheme
Ambassador
it IS important to keep a high single threaded score even when all cores are doing work and not lose 25-30% of the single threaded score if other cores do some work as well.
The first part of your statement doesn't align with the last. You quote the clockspeed drop from 5950X when all cores are loaded, and then go on to say "...if other cores do some work as well."

But, a situation where all other cores have a sustained load of the type that would drop clocks like what's shown in that plot is very different from a situation where "other cores do some work as well". The latter part sounds more like a 4-8 core loading scenario, where the 5950X lost only 6% to 9% of frequency per-core.
 

bit_user

Polypheme
Ambassador
Actually, in this case there is no difference. The KF just lacks an iGPU. You probably confused it with the KS? That one is actually slightly better.
All I know is that I tried switching it and the scores changed slightly. Not sure if that's a statistical anomally or something else, but the KF seemed to perform a little better.

Click Terry's link & try it yourself!
 

KyaraM

Admirable
All I know is that I tried switching it and the scores changed slightly. Not sure if that's a statistical anomally or something else, but the KF seemed to perform a little better.

Click Terry's link & try it yourself!
Honestly, that looks like it is related to there just being more data points for the 12900k than the 12900ks. That makes it more robust to diffences in individual results. My 12700k also fluctuates between 118 and 119 rating test to test, and sometimes scores only 117. Also, afaik the kf is often used in prebuilts. Due to bad optimization and cooling in those systems, CPUs can often underperform compared to their peers, which would lead to lower scores. But again, even the same CPU can have a certain range in values, so such small differences are essentially measurement variability.
https://cpu.userbenchmark.com/Compare/Intel-Core-i9-12900K-vs-Intel-Core-i9-12900KF/4118vsm1685583

As you can see, they are even equally rated. Because it really doesn't matter.