News Ryzen 5 7600X Beats i9-12900K by 22% in New Single-Core Benchmarks

[Admin]

AMD's upcoming Ryzen 5 7600X Zen 4 processor outperforms the Core i9-12900K in a new single-core benchmark.

Ryzen 5 7600X Beats i9-12900K by 22% in New Single-Core Benchmarks : Read more

 

[spongiemaster]

This has already been reported on other sites and the consensus seems to be that the single core result is either wrong or it's utilizing AVX-512 which is throwing the result off.

 

[-Fran-]

And in UserBenchmark nonetheless, LOL.

That site is such a joke that I can't even take this bit of "good news" seriously. That whole thing is a meme-site.

Regards.

 

[Alvar "Miles" Udell]

UserBenchmark...Can TomsHardware have a ban on any article mentioning that useless, meaningless site?

 

[SkyBill40]

User Benchmark. The meme site of benchmarking with a HUGE bias toward Intel. Hilarious.

Regardless, if this is to be truthful... it's a pretty significant win in terms of gain over the old process/design. Makes me wonder just how far the top dogs of the stack will be in front of their blue team competition? I suppose we'll see soon enough.

 

[user7007]

AMD says zen 4 ha 8-10% more IPC. I think this result is either an outlier that only happens in this particular test and isn't true for most workloads or fake.

 

[Bikki]

Ryzen 5 5600x Multi-Core land at 1198, which is 7.68 times greater than Single-Core (156), while the chip only have 6 cores. One explanation for greater than linear scaling is they measure the threads not the cores - Single thread vs Multi thread. Just a confusing benchmark I suppose.

 

[escksu]

I am not sure how accurate are the results. However, it's entirely possible to estimate 7600x results to very high accuracy.

Zen4 is expected to have around 5-10% IPC boost over zen3.. so at the same clockspeed 7600x should be around 170. And then 5ghz boost clock. 5600x boost clock is 4.6ghz. If 7600x is 5ghz, that's 10% more. So it will be around 187.. if 5.5ghz, that's 20%, around 205.

So, for that single core result. 7600 has to be running at close to 6ghz which is obviously impossible.

 

[The Historical Fidelity]

I am not sure how accurate are the results. However, it's entirely possible to estimate 7600x results to very high accuracy.

Zen4 is expected to have around 5-10% IPC boost over zen3.. so at the same clockspeed 7600x should be around 170. And then 5ghz boost clock. 5600x boost clock is 4.6ghz. If 7600x is 5ghz, that's 10% more. So it will be around 187.. if 5.5ghz, that's 20%, around 205.

So, for that single core result. 7600 has to be running at close to 6ghz which is obviously impossible.

Talking to insiders, that 8-10% IPC improvement is intentionally an understatement by AMD. They would rather over-deliver than under-deliver. Plus remember this 8-10% IPC improvement number is just an average, meaning some tasks will benefit from architectural changes more than others.

 

[peachpuff]

Raptor lake will have double the e-cores, shouldn't amd just give us 2 extra cores? Not sure if they'll be able to compete with an 8 core 7800x to a 8+16 13900 in multi threaded stuff.

 

[TerryLaze]

Raptor lake will have double the e-cores, shouldn't amd just give us 2 extra cores?

How though?! They are locked into their ccd having 8 cores, they would have to use two ccd for the 5900x and three for the 5950x which would increase the price by a lot and would introduce the same lag between the ccd we know from the 5950x.
It's too late for a redesign, which would also be very expensive.

 

[-Fran-]

How though?! They are locked into their ccd having 8 cores, they would have to use two ccd for the 5900x and three for the 5950x which would increase the price by a lot and would introduce the same lag between the ccd we know from the 5950x.
It's too late for a redesign, which would also be very expensive.

Plus, to add a bit on what you're saying, they're now including an iGPU in the I/O die, so even less space for extra cores in the package. I wonder if AMD could even make a 10-core CCD without having to redesign the CCD itself around the IF/mesh. It would be cool though. Maybe Zen4-D (dense) would do something along those lines?

Regards.

 

[Friesiansam]

Some say Zen 4 is faster, some don't. Until we get some genuine, verifiable results from respected publications, I'll take all articles like this with a shed-load of salt.

 

[TerryLaze]

Some say Zen 4 is faster, some don't. Until we get some genuine, verifiable results from respected publications, I'll take all articles like this with a shed-load of salt.

Eh the way benchmarking is being done now it doesn't even matter anymore, they only measure the CPUs bandwidth and not the speed.
Who cares how well a single core does in an app that you will never run on only a single core? You only need the multithread result.
The only thing a single core result would matter for would be an app that only does run on a single thread and no review site does that anymore.
So no matter how many verifiable results from respected publications you get you will never know which core is faster.

It's like measuring the width of a door in a place with many doors, to trying to find out how tall a person can fit through one of the doors walking straight. It's futile.

 

[The Historical Fidelity]

Plus, to add a bit on what you're saying, they're now including an iGPU in the I/O die, so even less space for extra cores in the package. I wonder if AMD could even make a 10-core CCD without having to redesign the CCD itself around the IF/mesh. It would be cool though. Maybe Zen4-D (dense) would do something along those lines?

Regards.

I have been theorizing since I heard of zen 4-D a couple years ago that if AMD wanted to match core parity with Intel, they could easily put one zen 4 8 core/16 thread performance ccd and one zen 4-D 16 core/32 thread efficiency ccd on a 7950XD cpu. I think zen 4-D will be a big competitor to Intel E-cores since zen 4-D has SMT.

 

[peachpuff]

How though?! They are locked into their ccd having 8 cores, they would have to use two ccd for the 5900x and three for the 5950x which would increase the price by a lot and would introduce the same lag between the ccd we know from the 5950x.
It's too late for a redesign, which would also be very expensive.

You're overthinking this, simply have the 8 core replace the 6 core in the lineup. So amd will have the 8 core vs intels 6p+8e 13600, 12core vs intels 8p+e8 13700 and 16core vs 13900 8p+16e.
Save the 6 core for the $199 segment rather than $299 like the 5600x launched at.

 

[Sleepy_Hollowed]

One can hope this is the case. Not impossible considering RAM latency and IPC improvements, though I can't wait for release for real benchmarks.

It also seems that power-wise, this is a safer bet in the mobile space and for preventing throttling in smaller cases if those power numbers are accurate.

Hopefully the 170 W won't be hit unless being completely slammed.

 

[TerryLaze]

You're overthinking this, simply have the 8 core replace the 6 core in the lineup. So amd will have the 8 core vs intels 6p+8e 13600, 12core vs intels 8p+e8 13700 and 16core vs 13900 8p+16e.
Save the 6 core for the $199 segment rather than $299 like the 5600x launched at.

So you are telling me that AMD should take an income hit of $150 from $450 the 8core costs now and a $100 hit on the 6core.
Also the high end users will be really happy about the lower end getting more cores but not the high end.

 

[KyaraM]

One can hope this is the case. Not impossible considering RAM latency and IPC improvements, though I can't wait for release for real benchmarks.

It also seems that power-wise, this is a safer bet in the mobile space and for preventing throttling in smaller cases if those power numbers are accurate.

Hopefully the 170 W won't be hit unless being completely slammed.

230W. 170W is TDP, 230W is PPT, or actually highest power draw allowed. 170W being the highest is a huge misconception spawned by AMD completely messing up their own presentation.

 

[hannibal]

How though?! They are locked into their ccd having 8 cores, they would have to use two ccd for the 5900x and three for the 5950x which would increase the price by a lot and would introduce the same lag between the ccd we know from the 5950x.
It's too late for a redesign, which would also be very expensive.

And also... AMD most likely don´t need more cores at this moment. I would personally take new Intel without e cores than with e cores for what I do with computers. So amd Zen4 only having performance cores is just fine if it is fast enough and it seems to be fast enough. But these are really early, and also from site that is pure bull****, so let see the real release and the pricing of total system. CPU + memory + motherboard considering you can upgrade your CPU in AMD later, with intel... most like not so much. Intels advantage is that you still can use DDR4 with it, but based on benchmarks, next gen Intel CPU really is much faster with DDR5 so... I expect that both companies starts at even ground in memory part.

 

[ottonis]

A huge boost in single thread performance but only a tiny increase in multicore performance doesn't sound overly plausible. Just saying...

 

[vinay2070]

Some say Zen 4 is faster, some don't. Until we get some genuine, verifiable results from respected publications, I'll take all articles like this with a shed-load of salt.

Exactly, cant even believe greymon or kopite anymore, thier leaks change almost every week.

 

[shady28]

Way to go showing a bench you don't understand Tom's.

If you look at UserBenchmark, faster speed rated memory begats faster CPU performance numbers even in single core. It is not a latency sensitive score. Clearly it's using memory copy as part of the CPU score in some capacity.

The DDR5 Alder Lake rigs get a 5%-10% bump in single core vs DDR4 on this bench.

The 200 score for a 12900K is the average of all rigs - and the vast majority are DDR4 which is far slower in terms of memory transfer and score lower on this bench.

The 7600X was using DDR5-5600 in the test.

A 12900K is getting 215 in single core with DDR5-4800.

A 12900KS with DDR5-6400 is getting 222 single core.

So this is really more like +10% single core. Still impressive, but a long way from 22%.

 

[mitch074]

To the people saying that i5 has more cores... Remember that AMD's SMP is more efficient than Intel's HT, and that Intel's E-cores don't have HT. Also, AMD's cores are pretty much half the size of Intel's for a very close IPC - thread for thread, AMD processors are roughly comparable to Intel's current lineup, and there is a reason why today's fastest gaming CPU is Zen 3 based with a cr*pload of cache.
If Zen 4has 10% better IPC and 10% higher clock speeds than Zen3, that's 20% more performance per core right off the bat. If you add 50% more RAM throughput, that's an extra 5-30% CPU efficiency depending on task. Here you go, you get your 250 points on UserBenchmark, the Intel shill so shill it gave a better rank to a Celeron compared with a same-generation Xeon.

 

[spongiemaster]

To the people saying that i5 has more cores... Remember that AMD's SMP is more efficient than Intel's HT, and that Intel's E-cores don't have HT. Also, AMD's cores are pretty much half the size of Intel's for a very close IPC - thread for thread, AMD processors are roughly comparable to Intel's current lineup, and there is a reason why today's fastest gaming CPU is Zen 3 based with a cr*pload of cache.
If Zen 4has 10% better IPC and 10% higher clock speeds than Zen3, that's 20% more performance per core right off the bat. If you add 50% more RAM throughput, that's an extra 5-30% CPU efficiency depending on task. Here you go, you get your 250 points on UserBenchmark, the Intel shill so shill it gave a better rank to a Celeron compared with a same-generation Xeon.

A 13700k engineering sample has been overclocked to 6GHz using an AIO. Single tread CPU-Z score is 52% higher than a 5950X.

https://videocardz.com/newz/intel-c...-allegedly-scores-983-points-in-cpu-z-st-test