Still that MT score is nothing to write home about. 6950X can easily do 4.3GHz.
You do not know what that one can do because mind you, it is clearly a stock run, as you can tell from looking at couple of things.Still that MT score is nothing to write home about. 6950X can easily do 4.3GHz.
And I would strongly suggest certain folks to stop looking at screens with a ton of unknown information and actually use hwbot if they want to paint very dry picture. Because in case of hwbot we can actually know parameters of system.I would strongly suggest at people with outdated knowledge refrain from posting about possible conspiracies when it is known that MT ratio is affected by background applications.
Science 101: when your data point comparison is of the same OOM as the error, it's pretty much useless.Btw I did not bother to reply to that nice pessimization of performance estimates. 10% better with +-7% error in my book means it could be also >15% faster.
Wrong. 6700K reference score is 474.Again, the important take away is that in CPU-Z ST bench 4.4Ghz Skylake-X equals 5Ghz Skylake-X. That is pretty much upper bound of IPC improvement.
Ok a quick eyeball's says the cores are quite a bit bigger assuming the DDR interfaces are about the same length, there is lots of wasted space on broadwell-E so if i had to guess we are looking at something like 400-450mm sq. If the DDR interfaces are smaller then this will be wrongAnyone going to guess the die size based of DDR4 memory interface relative to broadwell-E?
Wrong. 6700K reference score is 474.
Download version 1.79 and see for yourself what the reference score for the 6700K is.474*5/4.5=538.6666
He is right.I am wrong, okay, forgot for a minute that 6700k was the 4.2Ghz ST boost.
You are wrong still, 6700k ref score is http://valid.x86.fr/bench/1 455 landing us to 541 for 5Ghz "sibling"
Intel has really felt the AMD competition, who would have thought!
Looking at the line up, the i7-7820X looks to be where intel has really pushed to be as competitive as possible with AMD. 100$ more than the 1800x, but better clocks, better IPC, more PCI lanes etc. Not that bad of a deal.
Only problem is the X299 platform should be quite a bit more expensive..
You are wrong still, 6700k ref score is http://valid.x86.fr/bench/1 455 landing us to 541 for 5Ghz "sibling"
The only indication which may suggest that ITBM 3 is working is the reported clock speed.You do not know what that one can do because mind you, it is clearly a stock run, as you can tell from looking at couple of things.
Elmor has a result with the i9 7900X on hwbot and it is a 5.7GHz LN2. The Nearest 6950X I can find is a 5376 MHz LN2 and it does wPrime 1024m in 48.5s compared to 44.3 seconds for the 5700MHz 7900X.And I would strongly suggest certain folks to stop looking at screens with a ton of unknown information and actually use hwbot if they want to paint very dry picture. Because in case of hwbot we can actually know parameters of system.
Turbo boost at all working is the sign that you can't know what clocks MT score was done at. You can only know it was above base clock by unknown number.The only indication which may suggest that ITBM 3 is working is the reported clock speed.
Notice how i never said that hwbot results do not paint a very sad picture for Skylake-X. They do paint worse picture than any speculation on Intel's slides or chiphell leaks can muster.Elmor has a result with the i9 7900X on hwbot and it is a 5.7GHz LN2. The Nearest 6950X I can find is a 5376 MHz LN2 and it does wPrime 1024m in 48.5s compared to 44.3 seconds for the 5700MHz 7900X.
LOL. Gotta love these forums. Actually in that "530" score post I ran Cpu-Z on my 5Ghz 7700K machine (with TON of background apps) and got 530
Well, looks like someone has done it for me. Much appreciation goes to him.Download version 1.79 and see for yourself what the reference score for the 6700K is.
https://www.overclock3d.net/news/cp...x_and_kaby_lake-x_cpus_will_not_be_soldered/1
And they have to use toothpaste instead of soldering to the IHS with the price reduction.... oh dear...
To be fair, there is no problem with the paste, but let's hope that the gap is small enough it won't interfere with the cooling.
Is that an admission that Elmor's result shows that not much progress has been made in MT performance?Notice how i never said that hwbot results do not paint a very sad picture for Skylake-X. They do paint worse picture than any speculation on Intel's slides or chiphell leaks can muster.
ITBM 3 is different. It assigns threads to a single core that is allowed to boost beyond the regular turbo speed in light workloads.Turbo boost at all working is the sign that you can't know what clocks MT score was done at. You can only know it was above base clock by unknown number.
So from the die shots, what does the SKL-X die size look like? Around 3x compared to the KBL-X die?The actual video
Yes, backed with data i can be certain about. Unlike chiphell CPU-Zs.Is that an admission that Elmor's result shows that not much progress has been made in MT performance?
Thread assignment is a part of it but it is the irrelevant one. The relevant part is boosting one (two now) cores higher than ST turbo bin allows. And of course only working with power management (including turbo boost) enabled.ITBM 3 is different. It assigns threads to a single core that is allowed to boost beyond the regular turbo speed in light workloads.
At least 3. That's pretty large die tbh.So from the die shots, what does the SKL-X die size look like? Around 3x compared to the KBL-X die?
So from the die shots, what does the SKL-X die size look like? Around 3x compared to the KBL-X die?
der8auer showed the delid with a 6-C SKL-X, which means the LCC die. The HCC die would be even bigger, so there is a possibility that it *might* be soldered.My guess for the large die is at 341mm2 for now, with a range of 320-350mm2 but it's tricky to make a guess based on the die shot.
No idea about the 12 cores die, guess one can try to measure from that vid but yet to do so myself.
TBH i was thinking that it just blows that we don't have a 64 cores A75 at 3.6GHz (EDIT or a bit less for a sane TDP) as an alternative to Skylake X and Threadripper. Wouldn't be that big of a chip and a VM on each 8 cores cluster would be way fun.
Paying 5% more to SiLo for delidding it? Making up the issues out of no issues, you know.Reports that Skylake-X will be TIM-based are starting to arrive. There goes all that OC "headroom." Wonder how the folks planning to buy based on supposed 4.5 GHz OC are feeling right now.
Or you know, people like their processors to be under warranty, especially their $2000 processors. AMD is going to be launching "ThreadRipper" 16C SKUs; they're going to be priced at the $1k point; and they're going to be soldered with full warranty.Paying 5% more to SiLo for delidding it? Making up the issues out of no issues, you know.
People who like their processors to be under warranty do not overclock. Applies to AMD as well.Or you know, people like their processors to be under warranty, especially their $2000 processors.
You and I both know that overclocking does not void warranty unless you admit to it. Go ahead and try to deny delidding your processor.People who like their processors to be under warranty do not overclock. Applies to AMD as well.