Der8auer hits 5.0GHz on i9-7900X with AIO water cooling

[spaniel53]

 

He hit 4.8GHz without delidding! Only 200MHz gain after delidding. Looks like the excess adhesive issue has been resolved. Very exciting results. If these results are repeatable, people will likely be changing their tunes about Intel. I can't wait to hear how far the 6/8-core models can be pushed.

[Atmos]

1 minute ago, spaniel53 said:

If these results are repeatable, people will likely be changing their tunes about Intel.

After their attitude this last nearly decade and their complete and utter disaster of a launch with this newest series, I can confidently say that won't happen.

 

Intel has been digging this hole for a long time now, and now we dont have to keep holding on to their hand to stop them from slipping down it. AMD has come back from vacation and is giving us exactly what we've been demanding now for years, more cores on the mainstream platform.

[Benasurus]

I doubt this is representative. Intel most likely gave him the best silicon they could get rather than risk the silicon lottery.

[aGl1tch]

That's impressive, sure but Intel dun goofed with their pricing and this whole X299 platform is just a mess. 
I'm giving my money to AMD the next time I upgrade my CPU/Mobo, glad I only went for a budget B150 build when I did, screw Intel imo

[Gofspar]

2 minutes ago, Benasurus said:

I doubt this is representative. Intel most likely gave him the best silicon they could get rather than risk the silicon lottery.

nop, most of them are doing this,

people are already getting retail chips.

I know a few people who have had them for like 4 weeks now.

[tlink]

3 minutes ago, Gofspar said:

nop, most of them are doing this,

people are already getting retail chips.

I know a few people who have had them for like 4 weeks now.

most of them are hitting 5ghz?

[spaniel53]

Yes, Der8auer in the video even states that his was "pretested", or one of the better Bins. However, I would guess that even lower binned chips will likely reach 4.5GHz. I also agree that X299 is still horribly priced, but for those who want a 6/8/10/12 AND high single core performance, Intel is their choice (albeit the cost per performance is likely worse, especially concerning increased cost of the X299 chipset).

[Gofspar]

3 minutes ago, tlink said:

most of them are hitting 5ghz?

4.6-5

 

if you're unlucky or have potato cooler, probably 4.5

it's still a lottery.

[LukaH]

1 hour ago, spaniel53 said:

If these results are repeatable, people will likely be changing their tunes about Intel.

yeh right. just wait until intel tells you not to overclock

i mean the OC is impressive he didn't even have to use LN2, but i doubt an impressive OC gain will make people say. "well the platform is a mess but who cares OC GAINZZZ 2 D M4X" 

[Gofspar]

3 minutes ago, spaniel53 said:

Yes, Der8auer in the video even states that his was "pretested", or one of the better Bins. However, I would guess that even lower binned chips will likely reach 4.5GHz. I also agree that X299 is still horribly priced, but for those who want a 6/8/10/12 AND high single core performance, Intel is their choice (albeit the cost per performance is likely worse, especially concerning increased cost of the X299 chipset).

he sells binned chips himself over on https://www.caseking.de/

[Atmos]

5 minutes ago, spaniel53 said:

Yes, Der8auer in the video even states that his was "pretested", or one of the better Bins. However, I would guess that even lower binned chips will likely reach 4.5GHz. I also agree that X299 is still horribly priced, but for those who want a 6/8/10/12 AND high single core performance, Intel is their choice (albeit the cost per performance is likely worse, especially concerning increased cost of the X299 chipset).

You, realize that AMD's current offerings have barely less IPC than intels... right?

"High  single core performance" both clock per clock have very similar performance lol... This newest rehash is marginally faster aye, but not 500$ more performance.

[spaniel53]

2 minutes ago, LukaH said:

yeh right. just wait until intel tells you not to overclock

i mean the OC is impressive he didn't even have to use LN2, but i doubt an impressive OC gain will make people say. "well the platform is a mess but who cares OC GAINZZZ 2 D M4X" 

True, I agree. I suppose this platform will mostly appeal to those who want both 6+ cores with the best single core performance, even if it is a higher cost per performance. I think a good amount of enthusiasts would be willing to pay more, if only to have "the best performance".

[Valentyn]

A delid with liquid metal and it still hits 88-90 degrees under Stress test. That's kinda hot for the effort on a £1000 CPU. At least delidding with his new tool looks much easier than previously.

Will be interesting to see what frequency and temps reviews, and consumers can manage without a liquid metal delid.

 

Would have been great if he did some AIO overclocking before the delid as well, so we can have some sort of baseline for temps and max overclock before he delidded.

[LukaH]

1 hour ago, spaniel53 said:

True, I agree. I suppose this platform will mostly appeal to those who want both 6+ cores with the best single core performance, even if it is a higher cost per performance. I think a good amount of enthusiasts would be willing to pay more, if only to have "the best performance".

For me the X series chipset shouldn't even advertise single core performance it should be treated more as a side note. Especially with intel since it is (was) the only platform where you can get more than 4 cores aka "megatasking". 

[Enderman]

10 minutes ago, LukaH said:

For me the X series chipset shouldn't even advertise single core performance it should be treated more as a side note. Especially with intel since it is (was) the only platform where you can get more than 4 cores aka "megatasking". 

What?

Nobody is advertising single core performance.

It is common knowledge that Intel CPUs have better core performance than AMD.

It has been like this for a decade.

[LukaH]

1 hour ago, Enderman said:

What?

Nobody is advertising single core performance.

It is common knowledge that Intel CPUs have better core performance than AMD.

It has been like this for a decade.

yes this was a reply to someone who was emphasizing on single core performance.

[Enderman]

1 minute ago, LukaH said:

yes this was a reply to someone who was emphasizing on single core performance.

Someone talking about core performance  =/= the manufacturer advertising single core performance.

[raphidy]

27 minutes ago, huilun02 said:

...on all cores? 

Yeah, if you watched the video.

 

Those are insane clocks for a 10C chip.

[done12many2]

19 minutes ago, raphidy said:

Yeah, if you watched the video.

 

Those are insane clocks for a 10C chip.

 

You bet your ass it is.  Anything about 4.5 would be incredible with that many cores.  Multi-threaded performance while maintaining very high single-threaded performance.  

 

I delid chips.  It's just what I do.  I also run a very large water loop.  You hit the nail on the head.  Insane clocks.

 

That was tested using Prime95 non-AVX.  Imagine what it will do with the weak ass AIDA64 that everyone on LTT uses?  

[Valentyn]

1 minute ago, done12many2 said:

 

You bet your ass it is.  Anything about 4.5 would be incredible with that many cores.  Multi-threaded performance while maintaining very high single-threaded performance.  

 

I delid chips.  It's just what I do.  I also run a very large water loop.  You hit the nail on the head.  Insane clocks.

I'm really looking forward to reviewers and users overclocking it without a delid. If it can hit 4.2-4.5Ghz on an AIO without a delid under stress it's going to be great.

 

This actually resolves a lot of the issues I originally had, as it seems the delid knocked off around 10 degrees. Much better than Kabylake's TIM application.

[done12many2]

1 minute ago, Valentyn said:

I'm really looking forward to reviewers and users overclocking it without a delid. If it can hit 4.2-4.5Ghz on an AIO without a delid under stress it's going to be great.

 

This actually resolves a lot of the issues I originally had, as it seems the delid knocked off around 10 degrees. Much better than Kabylake's TIM application.

 

I actually edited right before you quoted me.

 

"That was tested using Prime95 non-AVX.  Imagine what it will do with the weak ass AIDA64 that everyone on LTT uses?  "

 

That's to say that at higher clocks, AIDA64 and RealBench stress testing will be a great deal lower in temps.

[PCGuy_5960]

9 minutes ago, done12many2 said:

You bet your ass it is.  Anything about 4.5 would be incredible with that many cores.  Multi-threaded performance while maintaining very high single-threaded performance.  

 

I delid chips.  It's just what I do.  I also run a very large water loop.  You hit the nail on the head.  Insane clocks.

 

That was tested using Prime95 non-AVX.  Imagine what it will do with the weak ass AIDA64 that everyone on LTT uses?  

If you get a good chip you should be able to get 5.2-5.3GHz with a custom loop, which is INSANE! Intel is hitting AMD where it hurts

[done12many2]

2 hours ago, LukaH said:

For me the X series chipset shouldn't even advertise single core performance it should be treated more as a side note. Especially with intel since it is (was) the only platform where you can get more than 4 cores aka "megatasking". 

1 hour ago, LukaH said:

yes this was a reply to someone who was emphasizing on single core performance.

 

Regardless of how many cores a chip has, single-threaded performance is still very relevant to consumers.  After all, 90% of the stuff that we actually do throughout a day is single-threaded.  While we may use our chips for multi-threaded work loads, that doesn't mean that's all it does.  

 

Why would you want a chip that is great in multi-threading only, but sucks in single-threaded?

[Valentyn]

6 minutes ago, done12many2 said:

 

I actually edited right before you quoted me.

 

"That was tested using Prime95 non-AVX.  Imagine what it will do with the weak ass AIDA64 that everyone on LTT uses?  "

 

That's to say that at higher clocks, AIDA64 and RealBench stress testing will be a great deal lower in temps.

 

Ah, I see. The main issue is Handbrake and other video encoding is using AVX more and more, so I'll definitely need to wait for other tests. Also AIDA64 actually has an FPU option to stress more. No idea why many don't test that, guess it's because AVX usage is still rather niché.

Although I'm still happy the TIM application isn't a 20+ degree difference, although he mentioned it was a tested chip before be overclocked, so hopefully it wasn't the top of golden samples.

[PCGuy_5960]

Just now, Valentyn said:

Although I'm still happy the TIM application isn't a 20+ degree difference, although he mentioned it was a tested chip before be overclocked, so hopefully it wasn't the top of golden samples.

Doubt it, another guy was able to get 4.5GHz with 1.15V so it doesn't look like this is a binned golden sample