Intel's 13th Gen Raptor Lake ES CPU gets Benchmarked

[TheLostSwede]

Just hours ago a CPU-Z screenshot of an Intel Raptor Lake ES CPU appeared and the same CPU now appears to have been put through a full battery of benchmark tests, courtesy of Expreview. This upcoming 13th gen Core CPU from Intel is limited to a maximum clock speed of 3.8 GHz and as such, was tested against a Core i9-12900K that was clocked at the same speed, for a fair comparison. Both CPUs were used with an unknown Z690 motherboard, 32 GB of DDR5 5200 MHz memory with unknown timings and a GeForce RTX 3090 Founders Edition graphics card. According to Expreview, the 13th gen CPU is on average around 20 percent faster than the 12th gen CPU, although the extra eight E-Cores might have something to do with that in certain benchmarks.

In Sisoft Sandra 2021 the ES sample is as much as 51.5 percent faster in the double precision floating point test, which is the extreme outlier, but it's ahead by around 15-25 percent in most of the other tests. In several other tests, it's ahead by anything from as little as less than three percent to as much as 25 percent, with more multithreaded types of benchmarks seeing the largest gains, as expected. However, in some of the single threaded tests, Alder Lake is edging out Raptor Lake by 10 percent or more, for example in Pov-Ray and Cinebench. Most of the game tests favour Intel's 12th gen over the 13th gen ES sample, although it's possible that the limited clock speeds are holding back the Raptor Lake CPU. The two are either neck in neck or Alder Lake being ahead with anything from a couple of percent to almost nine percent. Keep in mind that it's still early days and everything from UEFI support to drivers will be improved before Raptor Lake launches later this year. There's also the limited clock speed which is likely to play a significant role in the final performance as well, but this does at least provide a first taste of what's to come. Head over to Expreview for their full set of benchmarks.



View at TechPowerUp Main Site | Source

 

[ppn]

10% IPC loss could mean that raptor has slighlty relaxed latencies in order to clock higher, similar to sandy and ivy bridge, Ivy had to clock 100Mhz higher for same results or something, it was a long time ago.

 

[Space Lynx]

10% IPC loss could mean that raptor has slighlty relaxed latencies in order to clock higher, similar to sandy and ivy bridge, Ivy had to clock 100Mhz higher for same results or something, it was a long time ago.

Surely Intel wouldn't release something with a lower single thread performance?

If that is the case, I think I would prefer 12th gen Intel or hold out for 7000 series 3D cache from AMD

all I do is game, so the other crap doesn't interest me

 

[TheLostSwede]

Surely Intel wouldn't release something with a lower single thread performance?

If that is the case, I think I would prefer 12th gen Intel or hold out for 7000 series 3D cache from AMD

all I do is game, so the other crap doesn't interest me

No, that's not likely, as it should be compensated by higher clock speeds.

 

[rnt0]

10% IPC loss could mean that raptor has slighlty relaxed latencies in order to clock higher, similar to sandy and ivy bridge, Ivy had to clock 100Mhz higher for same results or something, it was a long time ago.

lol @ maths
st performance isnt purely ipc, its an es sample @ 65w hence the low st performance
they wont release a cpu with worse ipc

 

[Chane]

Looks like nice gains. I'm in the AMD camp right now, but Intel's got me intrigued again. I'm not surprised with ST being lower in a couple tests considering it's an ES chip.

 

[thegnome]

Doesn't sound very good, they said improved P-Core ipc, doubling of e-cores yet barely faster in MT.

 

[TheLostSwede]

Doesn't sound very good, they said improved P-Core ipc, doubling of e-cores yet barely faster in MT.

Keep in mind that this is an ES sample, they tend to be pretty "crap" compared to QS samples and might not reflect that actual performance of what will be sold in retail.

Looks like nice gains. I'm in the AMD camp right now, but Intel's got me intrigued again. I'm not surprised with ST being lower in a couple tests considering it's an ES chip.

It should be a close fight for the performance crown this time around.

 

[InVasMani]

Definately going to take a overall look at what both camps offer. I really see DDR4 and PCIE 5.0 as wild cards though that could make or break my decision.

 

[aQi]

Seems too early for a leak but on time considering AM5 socket debut.

 

[maxfly]

Way to early to give this any real credence. The fact that an es chip hangs with the 12900k @65w is a good sign that it will compete with am4 (hopefully). In a couple of months when there are some actual retail chips floating around, that will tell us whats what. Still, its nice to see something.

 

[Steevo]

Intel,’what you doin

 

[Minus Infinity]

I hope the ES is crap, because it's worrying to see RL losing at all to AL at the same clocks. I hope Intel isn't just going to try and clock the crap out of RL to try and beat Zen 4 at any power draw.

Look it won't matter much at all what you buy, both systems will have plenty of performance and I'm sure Intel will win a few benchmarks and AMD will win a few, but I know Zen 4 will dominate for energy efficiency and with 3D cache coming earlier than expected, they will reign supreme. Pretty sure my cpu will be a 7900X. I'm really waiting for Meteor Lake/Arrow Lake from Intel.

 

[AlwaysHope]

Too early to speculate about raptor lake performance for this, mainly for 2 main reasons;
1. Unkown Z690 motherboard & thus an early release bios to support raptor lake, keeping in mind with reference to the last generation that rocket lake was optimised for Z590 motherboards & I would not be surprised in the slightest if this is the same similar case this time around with unreleased Z790 boards.
2. Unkown timings on the DDR5 ram. Memory timings regardless of DRAM generation can & does have an effect on most benchmarks.

But that won't stop the commenters "speculating".

 

[matar]

10% IPC loss could mean that raptor has slighlty relaxed latencies in order to clock higher, similar to sandy and ivy bridge, Ivy had to clock 100Mhz higher for same results or something, it was a long time ago.

(Long time ago) Sandy Bridge is intel's last true leap of magic and wizardry.
I still have my X79 PC with my i7-3930k @4.3GHZ 4x8 32GB 1600MHZ 24/7 no bios tweaks all Auto on AIO cooler since 2012 with my Dell 3008 30" 2560x1600 GPUs & Raid SSDs only thing i upgraded is to GTX 970 back in 2015 and kept one of the 560Ti 2GB for PhysX
Just last year i bulit a new PC with an i9-9900KF 32GB 3200MHZ RTX 3060TI FTW3 Ultra , so the i7-3930k served me 10 years and still my kids use it and me sometimes.

 

[ravenhold]

I'm running Ivy Bridge with 22nm architecture. I will upgrade next year only I want to use more energy efficient product.

 

[Daven]

(Long time ago) Sandy Bridge is intel's last true leap of magic and wizardry.
I still have my X79 PC with my i7-3930k @4.3GHZ 4x8 32GB 1600MHZ 24/7 no bios tweaks all Auto on AIO cooler since 2012 with my Dell 3008 30" 2560x1600 GPUs & Raid SSDs only thing i upgraded is to GTX 970 back in 2015 and kept one of the 560Ti 2GB for PhysX
Just last year i bulit a new PC with an i9-9900KF 32GB 3200MHZ RTX 3060TI FTW3 Ultra , so the i7-3930k served me 10 years and still my kids use it and me sometimes.

This is a misconception created by those who want excuses not to upgarde. Alder Lake is many times faster than Sandy Bridge. Sandy Bridge is out of date tech that should be avoided if you want to do anything with a computer besides email and web browsing.

 

[matar]

This is a misconception created by those who want excuses not to upgarde. Alder Lake is many times faster than Sandy Bridge. Sandy Bridge is out of date tech that should be avoided if you want to do anything with a computer besides email and web browsing.

lol

 

[efikkan]

lol @ maths
st performance isnt purely ipc, its an es sample @ 65w hence the low st performance
they wont release a cpu with worse ipc

Contrary to popular belief, IPC isn't performance at all, it's instructions per clock. But it is one of the main driving factors behind ST performance, which is why IPC is very important.
I doubt Raptor Lake will be any worse than Alder Lake, but I wouldn't expect it to be much better either. It's a minor architectural overhaul, so we should expect small gains overall. But as with any change, there could be outliers where we see performance regressions too.

But overall, this is a benchmark of an ES sample in an unknown state. Even the reported clock speeds are usually intentionally incorrect, plus certain features can be disabled in firmware, so I wouldn't care about benchmarks until we see QS samples or final samples.

Too early to speculate about raptor lake performance for this, mainly for 2 main reasons;
1. Unkown Z690 motherboard & thus an early release bios to support raptor lake, keeping in mind with reference to the last generation that rocket lake was optimised for Z590 motherboards & I would not be surprised in the slightest if this is the same similar case this time around with unreleased Z790 boards.

BIOS shouldn't directly affect CPU performance.

 

[Mussels]

Surely Intel wouldn't release something with a lower single thread performance?

If that is the case, I think I would prefer 12th gen Intel or hold out for 7000 series 3D cache from AMD

all I do is game, so the other crap doesn't interest me

11th gen says they're always release something, if its a good idea or not
the 11900K coming out with two less cores, less cache and lower base clock speeds is something that everyone hopes they never do again, but it's been done before so it's a genuine concern


I fear that Intel will take the easy path and just add more e-cores, and focus on fixing them like AMD did with tweaking inter-CCX performance and latencies. If they've got leading ST performance even by the tiniest fraction, they can throw on E-cores and win MT benches and call it a day.

(Long time ago) Sandy Bridge is intel's last true leap of magic and wizardry.
I still have my X79 PC with my i7-3930k @4.3GHZ 4x8 32GB 1600MHZ 24/7 no bios tweaks all Auto on AIO cooler since 2012 with my Dell 3008 30" 2560x1600 GPUs & Raid SSDs only thing i upgraded is to GTX 970 back in 2015 and kept one of the 560Ti 2GB for PhysX
Just last year i bulit a new PC with an i9-9900KF 32GB 3200MHZ RTX 3060TI FTW3 Ultra , so the i7-3930k served me 10 years and still my kids use it and me sometimes.

Ivy bridge was a great leap forward, even if it just removed some artificial caps sandy bridge had.
My 2500K was hard locked to 1866Mhz on the ram, but swapping to my 3770k unlocked 2400Mhz (and overclocking above that) on the same board and RAM

I love the generations where things CHANGE.

Sandy bridge gave us great performance and easy overclocking.
Ivy bridge added insane ram speeds to that platform, and seriously - DDR3 2400 really, REALLY kicked the performance higher on those chips with later-generation GPU's.

Zen 1 changed the 4 cores + SMT formula and shook the entire tech industry (look at epyc/threadripper, intel going from four cores to 20)
Zen 3 changed the expectations for power efficiency and multi threaded performance - the 'perfected' multi CCX tech at its best (so far)
The5800x3D shook things again with game-changing gaming performance, although it's limited to a 'low' 8 cores at present

Alder lake isn't a favorite of mine, but it sure as hell changed things as much as Zen 1 did - instead of multiple matching CCX designs, they went big.LITTLE and forced a paradigm shift in operating systems we haven't truly felt yet.
This is really, really bordering on the ability to slot in a PCI-E CPU like the old co-processor days since windows can now handle different CPU architectures with different instruction sets in the same PC.

 

[efikkan]

11th gen says they're always release something, if its a good idea or not
the 11900K coming out with two less cores, less cache and lower base clock speeds is something that everyone hopes they never do again, but it's been done before so it's a genuine concern

Whether 10th gen was a good idea or not, is subjective.
But the fact remains that i9-11900K outperformed i9-10900K both in average, and in most productive workloads and gaming, so overall it's a faster product. Some like you are fixated by the core count, but you shouldn't, as the real thing that matters is performance. If AMD tomorrow released a quad core which beat all other mainstream CPUs, wouldn't a rational person buy that over a CPU with more cores?
As for cache, don't get fixated by that either. Cache sizes have varied a lot over the last 30 years, but other things beyond cache size matters, like bank configuration, latency and bandwidth.
And to be completely pedantic, claiming there is less cache is also quite misleading, as both L1 and L2 is significantly larger on 11th gen, and each cache line in L1/L2 is much more valuable than a cache line in L3. While the total L3 is a bit smaller, there is also fewer cores competing over it, so overall there is not much difference.

I fear that Intel will take the easy path and just add more e-cores, and focus on fixing them like AMD did with tweaking inter-CCX performance and latencies. If they've got leading ST performance even by the tiniest fraction, they can throw on E-cores and win MT benches and call it a day.

I fear that too. And they probably will, as the primary motivation for bringing E-cores to desktop is for marketing reasons. Specs like clock speed, core count and synthetic benchmarks is what sells, both in the OEM market and to most custom builders.

 

[blu3dragon]

Whether 10th gen was a good idea or not, is subjective.
But the fact remains that i9-11900K outperformed i9-10900K both in average, and in most productive workloads and gaming, so overall it's a faster product. Some like you are fixated by the core count, but you shouldn't, as the real thing that matters is performance. If AMD tomorrow released a quad core which beat all other mainstream CPUs, wouldn't a rational person buy that over a CPU with more cores?
As for cache, don't get fixated by that either. Cache sizes have varied a lot over the last 30 years, but other things beyond cache size matters, like bank configuration, latency and bandwidth.
And to be completely pedantic, claiming there is less cache is also quite misleading, as both L1 and L2 is significantly larger on 11th gen, and each cache line in L1/L2 is much more valuable than a cache line in L3. While the total L3 is a bit smaller, there is also fewer cores competing over it, so overall there is not much difference.


I fear that too. And they probably will, as the primary motivation for bringing E-cores to desktop is for marketing reasons. Specs like clock speed, core count and synthetic benchmarks is what sells, both in the OEM market and to most custom builders.

Thos chart seems to show 10900k being slightly faster on average.

 

[atomsymbol]

The statement "51.5 percent faster in the double precision floating point test" doesn't correspond with the chart.

 

[Mussels]

Whether 10th gen was a good idea or not, is subjective.
But the fact remains that i9-11900K outperformed i9-10900K both in average, and in most productive workloads and gaming, so overall it's a faster product. Some like you are fixated by the core count, but you shouldn't, as the real thing that matters is performance. If AMD tomorrow released a quad core which beat all other mainstream CPUs, wouldn't a rational person buy that over a CPU with more cores?
As for cache, don't get fixated by that either. Cache sizes have varied a lot over the last 30 years, but other things beyond cache size matters, like bank configuration, latency and bandwidth.
And to be completely pedantic, claiming there is less cache is also quite misleading, as both L1 and L2 is significantly larger on 11th gen, and each cache line in L1/L2 is much more valuable than a cache line in L3. While the total L3 is a bit smaller, there is also fewer cores competing over it, so overall there is not much difference.


I fear that too. And they probably will, as the primary motivation for bringing E-cores to desktop is for marketing reasons. Specs like clock speed, core count and synthetic benchmarks is what sells, both in the OEM market and to most custom builders.

Wait didnt they do something odd with 9 and 10 as well as 10 and 11?

Anyway, point was that sometimes an individual product can go backwards, even if the lineup as a whole goes forwards... the 11900k received the infamous title of a 'waste of sand'

 

[wahdangun]

No, that's not likely, as it should be compensated by higher clock speeds.

just like P4 and bulldozer