Discussion Intel current and future Lakes & Rapids thread

[coercitiv]

Yeah, but for core IPC comparisons you want best SKU.

Then we better hope Intel sent them the 11900K for the actual review.

I'm seriously surprised more and more people are annoyed with the fact that this time around we got very good and interesting information upcoming products. This forum lives and breathes leaks and rumors, yet somehow managed to have an allergic reaction to a comprehensive leak. I wonder if the same moral/scientific standards would have applied if 10700K was a top performer instead.

Personally I blame Intel for not sending 11900K to Mindfactory. They made the wrong mistake. /s

 

[JoeRambo]

The 11700K leak drama? I don't care. If anything it properly represents "stock" 11700K performance and exposes outright Intel stupidity in the face of strong competition.

The real question is if Anandtech will retest 11900K.

 

[Zucker2k]

Yeah. It seems there's also talk, at least for the 11700k, that the uncore seems to be limited to 4GHz. If true, that's absolutely retarded. I think what most of us said about there being no difference between the 11700k and 11900k reached Intel, and they decided to draw a clear line between the two chips just so they can charge way more for the halo 11900k. Just retarded.

Guy in this thread says 4.5GHz. That wouldn't be stock though.

 

[Tarkin77]

https://twitter.com/i/web/status/1369429800218202119

 

[coercitiv]

Guy in this thread says 4.5GHz. That wouldn't be stock though.

We also have this as example for stock vs. overclock.

 

[Exist50]

https://twitter.com/i/web/status/1369429800218202119

I feel like posting Piednoel is just flamebait, or entertainment at best. Does anyone seriously believe he knows what he's talking about in the slightest?

 

[coercitiv]

The real question is if Anandtech will retest 11900K.

They will. Either to validate or to correct the current one, they need an official review.

 

[Zucker2k]

We also have this as example for stock vs. overclock.

That's laughable difference in that test.

 

[Dave2150]

Yeah, what that also means, is that Anandtech wasted a ton of effort on testing wrong "K" CPU in a mode that no enthusiast will run it on.

Still, 11700K is bad CPU, 1:1 mode goes up to 3733 currently, that is even worse scaling than on AMD chips.

I suspect Ian knew full well what was up, since he has the review notes/guides from Intel.

Conducting the 'review' in the way he did, generated an insane amount of traffic for Anandtech, something they haven't had since the good old days when Anand Lal Shimpi was running the show, when reviews were on time, accurate and of a high level of knowledge.

Seems like nothing more than a (successful) cash grab, though was successful for the wrong reasons, likely to further hurt credibility/PR in the long term.

 

[Timorous]

I suspect Ian knew full well what was up, since he has the review notes/guides from Intel.

Conducting the 'review' in the way he did, generated an insane amount of traffic for Anandtech, something they haven't had since the good old days when Anand Lal Shimpi was running the show, when reviews were on time, accurate and of a high level of knowledge.

Seems like nothing more than a (successful) cash grab, though was successful for the wrong reasons, likely to further hurt credibility/PR in the long term.

Hardware Luxx have their review and they tested 1:1 mode. Outside of Metro Exodus it shows a roughly 3% gain over the 10700K @ 1080p. With Metro exodus it is around 5.5%. Their latency graph puts the 11700K between 10th gen and 9th gen latency so to be honest, when cross checking the latency graph in Ian's review it could easily have been running 1:1 as well.

Metro does show that some games get large performance improvements over 10th gen so I would not be surprised if Intel were able to find a few games for their graphs and looking at some of them Metro is one of them.

 

[coercitiv]

Hardware Luxx have their review and they tested 1:1 mode.

I don't think they did, Andreas said he used 3200Mhz speed and "Auto" for Gear mode, which sounds like 1:2 based on the slide posted earlier by @Zucker2k .

 

[coercitiv]

Good news, new BIOS seems to improve memory performance.

https://twitter.com/i/web/status/1369605151414222849

 

[Timorous]

Good news, new BIOS seems to improve memory performance.

View attachment 40873

https://twitter.com/i/web/status/1369605151414222849

That L3 latency is what Ian got in his review though so no reason to think Ian was using a BIOS like the one on the left here.

Interesting that HardwareLuxx get >600 in CB R20. They are using a Corsair H150i Pro cooler though not sure how that compares to a TRUE copper with 2x 173 CFM fans on it. Also those timings are looking pretty good. Did they use C14 timings in the review?

EDIT: NB clock 3900 vs 3999?

 

[coercitiv]

Did they use C14 timings in the review?

I would assume so, the figures for the memory benchmark in the review are exactly the same as the ones in the AIDA screenshot on the left.

 

[Timorous]

I would assume so, the figures for the memory benchmark in the review are exactly the same as the ones in the AIDA screenshot on the left.

Looking at both reviews the numbers / relative performance are pretty much the same for the 5800X, 11700K, 10700K, 9900KS in both reviews. The gaming difference could be due to the different selection of games (and maybe the C14 ram although if across the board then that should net out). The only real outlier is the CB 20 single thread score.

 

[DrMrLordX]

The real question is if Anandtech will retest 11900K.

You mean the 11700k? Anandtech hasn't posted any numbers for the 11900k yet. If all they do is run the 11900k with defaults then it should have 1:1 mem speed as well.

I don't think they did, Andreas said he used 3200Mhz speed and "Auto" for Gear mode, which sounds like 1:2 based on the slide posted earlier by @Zucker2k .

Why would Intel introduce such bizarre IMC behavior?

 

[Zucker2k]

Why would Intel introduce such bizarre IMC behavior?

Segmentation aka greed. The irony is they're introducing RAM overclocking on B560

 

[.vodka]

....If this behavior is confirmed in the final reviews, then this is definitely more stupid for Intel than usual.

Why keep shooting themselves in the foot in such bizarre ways?


Now I'm intrigued for the rest of the reviews on the 30th.

 

[Timorous]

Been having a think as to why the 1T R20 results are so different between HardwareLuxx and Anandtech when the nT results and other benchmarks are all pretty similar to each other. A BIOS issue is an obvious candidate but would hotspot temps be different enough between a True Copper and a Corsair H150i Pro to restrict the max boost clock in the Anandtech review?

 

[Shivansps]

Any news on how the UHD 750 is performing? i really dont expecting much out of 32EUs.

 

[jpiniero]

Been having a think as to why the 1T R20 results are so different between HardwareLuxx and Anandtech when the nT results and other benchmarks are all pretty similar to each other. A BIOS issue is an obvious candidate but would hotspot temps be different enough between a True Copper and a Corsair H150i Pro to restrict the max boost clock in the Anandtech review?

R20 1T is bandwidth dependent. Both Tiger Lake and Cezanne can get scores over 600 but it seems like it needs 4266 memory for that. 3200 memory and it's a lot lower. This from NBC's reviews.

 

[Timorous]

R20 1T is bandwidth dependent. Both Tiger Lake and Cezanne can get scores over 600 but it seems like it needs 4266 memory for that. 3200 memory and it's a lot lower. This from NBC's reviews.

HardwareLuxx were using 3200 ram though so I don't think it is a memory issue. C14 vs C20 would not make that much difference in this test would it?

 

[jpiniero]

HardwareLuxx were using 3200 ram though so I don't think it is a memory issue. C14 vs C20 would not make that much difference in this test would it?

I suppose. The official JEDEC spec is 20 for 3200 DDR4 but going lower does not violate Intel's memory overclocking rules. It does look like HardwareLuxx was still using 1:2 since it was set to Auto. But I suppose that the board that HardwareLuxx was using 1:1 anyway despite being set to Auto while Ian's was using 1:2.

 

[Zucker2k]

So, going by that slide, anything below i9 11900k must be tested at DDR4 2933, otherwise you lose performance by having the IMC drop into 1/2 mode.

 

[tamz_msc]

Does the new BIOS actually result in improved scores in benchmarks, not just AIDA64? The AGESA 1.2.0.1 update from AMD improved L3 bandwidth reporting in AIDA64 but doesn't do anything else. Do we know that it's not the case here?