Discussion i7-11700K preliminary results

[GunsMadeAmericaFree]

early benchmarks

more benchmarks

Just thought I would provide these so folks could draw their own conclusions. Evidently the processors have been getting sold for a few days now in Germany.

 

[lobz]

My post was actually being sarcastic, though I appreciate this may not have been obvious. I couldn't care less about 1080P game FPS, as in my eyes anything below 4k is now a obsolete, retro only experience.

I game at 4k, on a 6700k, 3080, CX48 system, where I'd not be able to tell the difference between a 5800x, 10700k, 11700k or 11900k as games are mostly GPU limited at 4k.

I found a few games were my 6700k quad core, that I've had since August 2015, no longer copes, and I get FPS spikes from > 100 fps to 20 or 30. Shadow of the Tomb Raider's town scenes is one example, so many NPC's running around. I've not even tried Cyberpunk yet, due to waiting for a good CPU to be in stock and all the bugs.

My original plan was to get a 5900x, but in the UK they've not been in stock even once.

Sitting here with motherboard and new cooler ready to go, just need that 11900k to arrive

There isn't a single character in this post that I couldn't agree with.

 

[BFG10K]

There's a good supply of Ryzen CPUs @ MSRP according to this: https://hothardware.com/news/ryzen-7-5800x-cpu-retailers-amd-msrp

If Intel can't rely on AMD being out of stock, they'll have to undercut them on price.

 

[moonbogg]

There's a good supply of Ryzen CPUs @ MSRP according to this: https://hothardware.com/news/ryzen-7-5800x-cpu-retailers-amd-msrp

If Intel can't rely on AMD being out of stock, they'll have to undercut them on price.

That should be pretty easy considering the 5800X is the most expensive 8 core CPU since the time when Intel was a thing. I hope I get to see them charge $600 for the 11900K just for a good laugh. That would be some must-see entertainment. The talk of the town that would be.

 

[Jaskalas]

Intel is shooting themselves in the foot with these outright stupid power limits. There's no reason they need to...

Considering they had to backport to 14nm, I'd say they have a very good reason. They need that much power just to compete. Not only with AMD but with their own 10xx parts. I doubt they built toaster ovens without due cause.

 

[coercitiv]

the 5800X is the most expensive 8 core CPU since the time when Intel was a thing.

The 9900K was launched at $488, the 1800X was priced at $499.

 

[coercitiv]

They need that much power just to compete.

No they don't, they don't even get positive press coverage with these power limits.

Not only with AMD but with their own 10xx parts.

And know why they have a hard time competing with their own 10xx parts? Because they disabled power limits on 10xx parts as well. They set themselves up since years ago, since Coffee Lake hit the market. (although it became blatantly obvious with 9900KS and Comet Lake)

All they had to do to get the best of both worlds was enforcing proper stock TDP while also making it easy as pie to remove power limits for interested users. Remember the TURBO button in old PCs? Now they can do this in software. They could have made it trivial to adjust power and performance.

I guess we'll see how this pans out in a couple of weeks. Maybe the BIOS updates do fix inter-core latency and the product ends up being the gaming king, a much need saving grace, but anything less than Comet Lake gaming performance means the press will not be kind at all.

 

[moonbogg]

The 9900K was launched at $488, the 1800X was priced at $499.

I was remembering the actual price when the chip landed rather than msrp. $580+

Intel's Core i9-9900K Finally Available at Eye-Watering Prices

Intel's Core i9-9900K is now available -- for a certain definition of 'available.'

www.extremetech.com

 

[Sable]

There's a good supply of Ryzen CPUs @ MSRP according to this: https://hothardware.com/news/ryzen-7-5800x-cpu-retailers-amd-msrp

If Intel can't rely on AMD being out of stock, they'll have to undercut them on price.

Yup. In stock everywhere in the UK and for a reasonable (with UK/Europe shafting tax) £420.

 

[jpiniero]

If you look at the Photoscan test, the 11700K is faster than 10700K but only barely and is using way more power. The Perf/Watt of this thing is just bad even compared to Skylake.

 

[coercitiv]

It seems "the press" is quite upset with RKL-S ending up in reviews before the NDA lifts, and Anandtech is the obvious target.

 

[scineram]

This review has really brought out so many people on the IDF, its been incredible after reading so many forums. The full review for Rocket lake cpu's is going to melt the internet or cause heart attacks.

Well, they did warn that

Our email seemingly generated some excitement inside (and to our surprise, outside) Intel, but we received a response from Intel stating that they had no comment to offer.

 

[Kocicak]

I believe this is journalism as it should be.

BTW I belive that hungry journalists with open benches would welcome a little pan supplied with the new cpus, because the 300W is in sunny side up egg territory.

Just a remark about the claim that the CPU performance can improve with bios update. As Intel apparently was given a chance to get their statement published with the review and declined to comment, there are probably no significant performance improvements with bios update coming. Because Intel has no interest in their products being presented with lower than final performance. If they knew that the CPU will noticeably improve with the new bios, they would say so.

 

[DrMrLordX]

If you look at the Photoscan test, the 11700K is faster than 10700K but only barely and is using way more power. The Perf/Watt of this thing is just bad even compared to Skylake.

. . . which is why I'm thinking that a lot of people will be better off with an overclocked 10600k or 10700k versus this thing. Unless for some bizarre reason you need it for AVX512 in an application that can make proper use of it. Intel will be its own worst enemy unless they can get Comet Lake-S off the market, pronto. And you can't tell me that OEMs are excited about this product either!

It seems "the press" is quite upset with RKL-S ending up in reviews before the NDA lifts, and Anandtech is the obvious target.

Sour grapes. AT did their due diligence, left a paper trail, informed Intel, etc. If Intel isn't going to come out against them then it's okay for them to publish in my book.

@Kocicak

Time will tell. Some people have reported that Z590 boards are not ready for prime time. It may be more the board that's at fault than anything else? Hard to say what's really going on here.

 

[jpiniero]

Intel will be its own worst enemy unless they can get Comet Lake-S off the market, pronto.

If anything they are more likely to prioritize Comet Lake because it's also smaller.

 

[DrMrLordX]

If anything they are more likely to prioritize Comet Lake because it's also smaller.

I mean . . . I guess? It's a difficult situation for Intel regardless.

 

[jpiniero]

I mean . . . I guess? It's a difficult situation for Intel regardless.

OEMs might still be okay with Rocket Lake if they wanted the new IGP features. Or maybe they will just use Tiger Lake and just do BGA if supply is actually decent. Cezanne bascially doesn't exist so the competition is a lot less if you desire an IGP.

When Intel started talking about Alder Lake you had to figure that something was up.

 

[PingSpike]

The 9900K was launched at $488, the 1800X was priced at $499.

While that is true, in both of those cases there were more affordable 8 cores available in the lineup. 5000 series 8 core products start and end with the 5800x. The 5800x is a good chip, but its hardly a bargain.

 

[NTMBK]

And you can't tell me that OEMs are excited about this product either!

OEMs will probably run their motherboards with properly configured power limits. It's usually enthusiast boards that see this really insane power consumption, because they basically break Intel and AMD's power limiting mechanisms in order to win benchmarks.

This chip gives OEMs new "11th gen" and "Xe graphics" stickers to impress non-technical buyers, as well as a new microarchitecture with pretty good performance to impress technical buyers. And most importantly- Intel can actually supply them with high volumes of parts, while AMD is completely maxed out. I'm sure that HP and Dell will sell an absolute truckload of these parts, just like every other generation of Intel.

 

[PingSpike]

OEMs might still be okay with Rocket Lake if they wanted the new IGP features. Or maybe they will just use Tiger Lake and just do BGA if supply is actually decent. Cezanne bascially doesn't exist so the competition is a lot less if you desire an IGP.

When Intel started talking about Alder Lake you had to figure that something was up.

Rocketlake has a badly needed PCIe lane boost as well. Its certainly not something most people will care much about but it does have a bit of a niche there. Unfortunately, Zen has a very similar loadout so it really only gets them to parity.

 

[dullard]

They've also failed to provide the test setup details on the URL referenced in that slide.

Intel might have worse processors and crap cherry picking marketing. But they usually do pretty well at providing their benchmark details so that anyone can reproduce them.

• Go to the website on the slide: http://www.intel.com/PerformanceIndex
• Click the "CES 2021" link. https://edc.intel.com/content/www/us/en/products/performance/benchmarks/ces-2021/
• Scroll to "Claim Details- 11th Gen Intel Core S-series Demo"

Metro Exodus- In game benchmark, 1080p High settings. Processor: 11th Gen Intel® Core™ i9-11900K processor (RKL-S) PL1=125W TDP, 8C16T, Motherboard: MSI Z490 GODLIKE. MCE disabled, Bios Version: EC70I6, Memory: 32 GB (4x8) DDR4-3200 DDR4 SDRAM, Storage: Samsung 970 Evo Plus 1TB, Display Resolution: 1920x1080, OS: Microsoft Windows 10 Pro 20H2, Graphics card: EVGA RTX 3080 XC3, Graphics driver: 460.79

Versus

Processor: AMD Ryzen™ 9 5900X processor PL1=105W TDP, 12C24T, Motherboard: Production Asus Crosshair VIII Formula X570, Bios Version: 2702, Memory: 32 GB (4x8) DDR4-3200 DDR4 SDRAM, Storage: Samsung 970 Evo Plus 1TB, Display Resolution: 1920x1080, OS: Microsoft Windows 10 Pro 20H2, Graphics card: EVGA RTX 3080 XC3, Graphics driver: 460.79. Testing by Intel as of December 21, 2020

 

[Mopetar]

The 9900K was launched at $488, the 1800X was priced at $499.

In fairness to the 1800X it launched as a consumer desktop CPU at a time where anything more than 4 cores meant venturing into the HEDT space for Intel. An 8-core Broadwell-E was over $1000 and even the 8-core Skylake-X was more than $500.

The 1800X was the most expensive 8-core consumer desktop part, but the 1700X was only $400 for slightly lower clocks and the 1700 was $330 for 8-cores in a 65W thermal envelope. No one considered those prices expensive. We were all blown away by how cheap they were.

The industry was completely changed from that point. AMD basically stole fire from the gods as far as I'm concerned.

 

[TheELF]

I believe this is journalism as it should be.

BTW I belive that hungry journalists with open benches would welcome a little pan supplied with the new cpus, because the 300W is in sunny side up egg territory.

Just a remark about the claim that the CPU performance can improve with bios update. As Intel apparently was given a chance to get their statement published with the review and declined to comment, there are probably no significant performance improvements with bios update coming. Because Intel has no interest in their products being presented with lower than final performance. If they knew that the CPU will noticeably improve with the new bios, they would say so.

Anything intel could have said would have made them look defensive like they have something to hide or like being the sore loser.
They don't care if the press all measures the power in the worst case scenario and show this as being the only power state there is.
They know that any big client (that cares) runs their own tests on these chips before committing to buying loads of them.
Intel has seen the sales numbers for the 10th gen that had the same power draw with worse IPC.

 

[Justinus]

Anything intel could have said would have made them look defensive like they have something to hide or like being the sore loser.
They don't care if the press all measures the power in the worst case scenario and show this as being the only power state there is.
They know that any big client (that cares) runs their own tests on these chips before committing to buying loads of them.
Intel has seen the sales numbers for the 10th gen that had the same power draw with worse IPC.

If CapFrameX is correct and Intel has a microcode/bios update for z590 that improves latency and cache performance leading to rocketlake beating comet lake in games, Intel could have simply informed Ian there is a microcode update in the works and things could be much different.

For starters, Ian may have delayed the review until Intel could provide the important performance-changing bios. Or Ian could have reported that this performance is confirmed not final and will improve according to Intel. Or Ian could have left gaming benchmarks out pending receiving the new bios.

Intel choosing to say nothing knowing poor performance results will be published is a PR nightmare, if the performance figures are not accurate. You can already see the Intel marketing machine managing to get all kinds of attacks on Ian, his results, and his process by seemingly independent actors. I ask, why?

 

[Makaveli]

If CapFrameX is correct and Intel has a microcode/bios update for z590 that improves latency and cache performance leading to rocketlake beating comet lake in games, Intel could have simply informed Ian there is a microcode update in the works and things could be much different.

For starters, Ian may have delayed the review until Intel could provide the important performance-changing bios. Or Ian could have reported that this performance is confirmed not final and will improve according to Intel. Or Ian could have left gaming benchmarks out pending receiving the new bios.

Intel choosing to say nothing knowing poor performance results will be published is a PR nightmare, if the performance figures are not accurate. You can already see the Intel marketing machine managing to get all kinds of attacks on Ian, his results, and his process by seemingly independent actors. I ask, why?

It would be the first microcode update I've ever seen that is boosting performance essentially on the intel side, since the last few years have just been patching for Spectre and meltdown which have reduced performance.

 

[DrMrLordX]

OEMs will probably run their motherboards with properly configured power limits.

Sure, but remember that it probably has the same PL2 and tau values as Comet Lake. OEMs are notoriously cheap with their cooling, and even 5-10W increase in power consumption generation-over-generation is something that will not sit well with them if it results in a poor customer experience when combined with lackluster cooling solutions. Rocket Lake-S has demonstrated far too many instances where it requires more power for the same (or slightly better) performance core-per-core as Comet Lake-S, and that is not what OEMs want to see in their next gen products.

And most importantly- Intel can actually supply them with high volumes of parts, while AMD is completely maxed out. I'm sure that HP and Dell will sell an absolute truckload of these parts, just like every other generation of Intel.

I wasn't really bringing AMD into the picture. Yeah Rocket Lake brings with it a few new features, but that's going to be a bitter pill for them to swallow compared to previous gen Intel CPUs that aren't demonstrably worse.