If history repeats itself, i rather have a few more slower cores then fewer faster cores. I remember all the E8400 vs Q6600 threads. When next gen is in full swing people on a 9900K/9900KS might be wishing they sat on a 3900x. Could be wrong but i got a feeling i might be right.
If these are normal results for the 9900KS then there is really no need for Intel to make 10nm chips. It's not like the microarch changed, for this increase in efficiency they basically made a bigger improvement than most other companies new nodes. If that was the case Intel would be talking it up, so I am more inclined to attribute this to measurement difficulties or a lucky sample.
From our early analysis, watercooling is going to be a must with this processor. We registered temperatures in the 85-89C range during extended AVX stress tests at stock settings with the Corsair H115i cooler. However, it is noteworthy that the processor maintained a 5.0 GHz clock rate during those tests. Those stock temperatures dropped to the 60-64C range with our custom water cooling loop. After tuning to 5.2 GHz, temperatures hovered in the 77-81C range during an extended AVX stress test (custom loop).
If these are normal results for the 9900KS then there is really no need for Intel to make 10nm chips.
Exactly. I have a custom loop with dual 360 rads, and I use it to cool a 32 core 2990wx@3.8. It gets too hot@4.0. So for them to use the same cooling on an 8 core chip says a lot about its heat production, 4 times a 12 (or its it 14nm) 32 core chip ?Or exceptional cooling... it's well-known that cooling chips well reduces the voltage (and therefore, power consumption) required to achieve a target frequency. That's why the "CL" option for "Custom Loop" used less package power, though it's going to be offset by the additional power used by the pump and fans versus a standard air cooler.
From the article:
Two things to note:
1) Their 280mm AIO struggled to cool the chip
2) They used a custom loop with dual 360mm rads for the numbers in the graph
You forgot this part:Exactly. I have a custom loop with dual 360 rads, and I use it to cool a 32 core 2990wx@3.8. It gets too hot@4.0. So for them to use the same cooling on an 8 core chip says a lot about its heat production, 4 times a 12 (or its it 14nm) 32 core chip ?
That's excellent, anyway you slice it. You can compare those temps and clocks to AMD's entire 7nm range and they won't do as well.We registered temperatures in the 85-89C range during extended AVX stress tests at stock settings with the Corsair H115i cooler. However, it is noteworthy that the processor maintained a 5.0 GHz clock rate during those tests.
Your crapping on the chip in this thread certainly does not sound like a recommendation.
89c under load is excellent ? Not in my book.You forgot this part:
That's excellent, anyway you slice it. You can compare those temps and clocks to AMD's entire 7nm range and they won't do as well.
Try running your 2990WX at 4GHz with a Corsair H115i and let's see what temps you get.89c under load is excellent ? Not in my book.
Good gamer chip. Looks like it'll cost too much, though.
@Abwx
y-cruncher is really close to Prime95. It has pauses, but it does use AVX2 and AVX512 if your CPU support them. It can be really intense.
Not this oneObviously they are not using AVX2 at all since the 9900K use 130W, exactly what it drain in Cinebench R15, with AVX2 and Prime 95 it goes up to 185W despite a frequency offset.
Apparently this is a paid review by who you knows, their statement about AVX is indeed purposely vague as they do not state AVX2, only AVX, and for a reason...
The numbers for the 9900K, for CB and Prime :
AMD Ryzen 9 3900X, 7 3700X & 5 3600 im Test: Leistungsaufnahme, Temperatur, Effizienz
AMD Ryzen 3000 im Test: Leistungsaufnahme, Temperatur, Effizienz / Leistungsaufnahme im Leerlauf und unter Lastwww.computerbase.de
The 9900KS should display 9% higher deltas than the the K in Cinebench and about the same power in Prime given that there will be the same AVX2 offset.
It could wait until 7nm but like I said, I don't think Intel's done much special here. Rather there exists some other circumstance in their testing setup.Except for that whole density thing.
That's probably it.Or exceptional cooling... it's well-known that cooling chips well reduces the voltage (and therefore, power consumption) required to achieve a target frequency. That's why the "CL" option for "Custom Loop" used less package power, though it's going to be offset by the additional power used by the pump and fans versus a standard air cooler.
From the article:
Two things to note:
1) Their 280mm AIO struggled to cool the chip
2) They used a custom loop with dual 360mm rads for the numbers in the graph
Obviously they are not using AVX2 at all since the 9900K use 130W, exactly what it drain in Cinebench R15, with AVX2 and Prime 95 it goes up to 185W despite a frequency offset.
Not this one
Intel did something and regular 9900k consumes nearly the same as 5GHz ks
Look at the link they tested avx loads with aida
Ofc we need more p/reviews to validate
Now that is just a stupid statement. I have 4 times the cores. And 89c on a 8 core CPU ? My 12 core 3900x does 71c at 4.1 ghz, 100% load on a 240 mm aio. They have a 280mm AIO, and it only manages 89c ?Try running your 2990WX at 4GHz with a Corsair H115i and let's see what temps you get.
The power draw can vary from review to review depending on the mainboard and PL1/PL2 wattage settings in the bios.
The data is there, maybe one could, you know, let the reader make up his own mind, without hyperbolic commentary.How could you possibly recommend it without severe caveats?
We are giving our opinions of the data, thats what this forum is for. Data must be interpreted to make any sense.The data is there, maybe one could, you know, let the reader make up his own mind.
Well, Mark, if I stated my true "opinion" I am sure it would garner an immediate infraction. For a "politically correct" response, see my post above.We are giving our opinions of the data, thats what this forum is for. Data must be interpreted to make any sense.
If I stated my true "opinion" I would also get an infraction. But with all the evidence all users have posted here, it would seem that your opinion is in a quorum of 2, the OP and you, both Intel advocates.Well, Mark, if I stated my true "opinion" I am sure it would garner an immediate infraction. For a "politically correct" response, see my post above.
You're just blinded by your bias. You implied that the 9900KS needed dual rads to be cooled with this post:Now that is just a stupid statement. I have 4 times the cores. And 89c on a 8 core CPU ? My 12 core 3900x does 71c at 4.1 ghz, 100% load on a 240 mm aio. They have a 280mm AIO, and it only manages 89c ?
Its a HOT chip
I showed you that is not the case. It run extended AVX stress tests at 5GHz with a Corsair H115i cooler and only registered a high of 89c. Just to put things in perspective, I reversed their testing methodology by matching your 2990WX with the Corsair H115i cooler, and of course, you're triggered. You have to be consistent with your claims. Does the 9900KS running at 5GHz require the same cooling as a 2990WX running at 4.0GHz?I have a custom loop with dual 360 rads, and I use it to cool a 32 core 2990wx@3.8. It gets too hot@4.0. So for them to use the same cooling on an 8 core chip says a lot about its heat production, 4 times a 12 (or its it 14nm) 32 core chip ?
The 9900KS at 5.2GHz (non Custom Loop) consumes 177 watts, which leads me to think that the dual rad was unnecessary - something made abundantly clear by the fact that the H115i kept the chip under 90c even with extended AVX stress tests.That's probably it.
I doubt they can do 5GHz on one core, much less 8.it's not like they couldn't bin their 3900X to 5ghz, so, why?? Do they not have a marketing department?
Come on, they are not that far off.I doubt they can do 5GHz on one core, much less 12.
That's a single core that may not be doing any useful work at that speed, given what we have seen with AMD's remedy to the boosting saga in earlier releases. I'm sure @DigDog was talking about all core overclock, which the 9900KS will do.Come on, they are not that far off.
I mean the upcoming 3950X, a 16 core CPU, boosts up to 4.7 Ghz. That's right in line with Intel's upcoming Cascade Lake-X line. While agree that 5 Ghz helps Intel CPUs, it's not like AMD is right there in overall performance (regardless of base / boost speeds).
I doubt anyone who either buys a i9-9900KS or Ryzen 3900X is going to be that disappointed with their purchase as they are both great CPUs.