coolalerIntel Core i7-5960X CPU-Z

[csbin]

http://www.coolaler.com/showthread.php/315765

8c16t
Boost 3.3GHz

 

[OCGuy]

Am I seeing that correctly? Max boost @ .978V with all 16 logical cores enabled?

Even though that is a cherry-picked ES, one could imagine the headroom still available with the proper cooling solution.

Awesome stuff...

Anyone know the significance of the M0 stepping?

 

[mikk]

Voltage on unreleased hardware is unreliable, too many CPUz misreads.

 

[Homeles]

Anyone know the significance of the M0 stepping?

I'm sure it's not the answer that you're looking for, but...

It does mean it's undergone a helluva lot of revision. 13... yikes. Guess it's not as bad as Ivy Bridge (S1), but it's no wonder why those two were/are so late.

 

[Khato]

I'm sure it's not the answer that you're looking for, but...

It does mean it's undergone a helluva lot of revision. 13... yikes. Guess it's not as bad as Ivy Bridge (S1), but it's no wonder why those two were/are so late.

Actually it doesn't mean that any more. They haven't started at A for every SKU in a generation since SNB. The pattern appears to typically be that the lead product starts at A0 while the alternate configurations start somewhere else in the alphabet.

 

[kimmel]

I'm sure it's not the answer that you're looking for, but...

It does mean it's undergone a helluva lot of revision. 13... yikes. Guess it's not as bad as Ivy Bridge (S1), but it's no wonder why those two were/are so late.

That is assuming they started this die at A0.

 

[Pheesh]

each die line begins in a certain area of the alphabet, so would not presume tons of revisions. I think if you look at Ivy bridge it might have been M-step for the 10-8 core parts, R-step for 6-4 core?

 

[Rakehellion]

Double the cores, double the heat?

 

[Arachnotronic]

Double the cores, double the heat?

Double the cores, more than double the PCIe lanes, nearly triple the L3$, and over double the memory bandwidth for double the heat.

Seems like a pretty good trade-off actually.

 

[aigomorla]

Am I seeing that correctly? Max boost @ .978V with all 16 logical cores enabled?

Possibly yes, and possibly no...
Without more information on the test setup, its hard to verify.
Ie... is it a new platform?
Ie... is it a new board?

If you recall when i did gulftown, CPU-Z was displaying information correctly, however coretemp was all over the place trying to identify the last 2 physical cores.

 

[SAAA]

Only 3.3 turbo? I expected a much more aggresive scaling on the top bin part... maybe it's to further differentiate it from faster DCs quad?
If it's soldered there should be less overheat problems anyway.

 

[OCGuy]

Double the cores, double the heat?

Heat has to do with the amount of power being consumed.

 

[SAAA]

Heat has to do with the amount of power being consumed.

Indeed it's not going to consume more than any other Haswell chip when running the same amount of threads, still it makes me wonder if it isn't time for a more advanced overclock vs the regular "all cores at maximum speed POZZIBLE!". With this many cores wouldn't make more sense go the turbo bins route and put say +15 bins on single thread load, and less for any increase up to say +5/6 on full load?

 

[OCGuy]

Indeed it's not going to consume more than any other Haswell chip when running the same amount of threads, still it makes me wonder if it isn't time for a more advanced overclock vs the regular "all cores at maximum speed POZZIBLE!". With this many cores wouldn't make more sense go the turbo bins route and put say +15 bins on single thread load, and less for any increase up to say +5/6 on full load?

My mobo allows me to OC specific cores.


Using speedstep/base-clock/boost-clock for CPUs is foreign to me. I put constant voltage through my max tested stable clock at all times.

 

[SlowSpyder]

Indeed it's not going to consume more than any other Haswell chip when running the same amount of threads, still it makes me wonder if it isn't time for a more advanced overclock vs the regular "all cores at maximum speed POZZIBLE!". With this many cores wouldn't make more sense go the turbo bins route and put say +15 bins on single thread load, and less for any increase up to say +5/6 on full load?

When I bought my FX, that was my plan. But AMD's modules don't have their own power feeds separate from the rest of the CPU as far as I can tell. I figured having a single module as high as I could get it (say, ~5.5GHz, the rest of the CPU @ ~4.7GHz) would probably be a better trade off for power use and heat compared to overclocking all cores, and give as good of single threaded performance as you could otherwise get, while still providing really great multi threaded performance. If Intel allows for that kind of tweaking, I'd be curious to see how much more a single core or two can overclock compared to a whole CPU approach.

 

[SAAA]

Say, ~5.5GHz, the rest of the CPU @ ~4.7GHz.

You mean when under load, do you? That would be even better if possible than the bins route: imagine a game where the main thread is running in the highest clocked core and the rest in all the cores you have left, even at GHz less of speed. It should be the better tradeoff beside any big.Little or unsymmetrical cores, if they will ever happen in x86.

 

[YBS1]

If Intel allows for that kind of tweaking, I'd be curious to see how much more a single core or two can overclock compared to a whole CPU approach.

That may not go as well as you'd think. I'm sure it will vary somewhat with each individual chip, but using my 3930K as an example, it didn't seem to have any spectacular or dud cores. I found this out while trying to push up some single threaded benchmarking scores. All six are stable @5.1, but even with the other cores disabled not a single one of them was stable @5.2 within the voltage range I was willing to push.

 

[Shimmishim]

That makes me mad that they made the 5930 - 6 core instead of 8 like the 4960 and 4930 were both 6 core.

 

[RussianSensation]

That makes me mad that they made the 5930 - 6 core instead of 8 like the 4960 and 4930 were both 6 core.

If anything what they did makes more sense. Before, they were charging $1K for more or less extra cache a small bump in clocks over the 4930. In terms of manufacturing, it doesn't cost them much more to make the 4960 vs. 4930, but how many enthusiasts had any incentive to buy the 4960X? Intel was losing out on a lot of profits. Now by moving the 8-core exclusively into the X realm this finally justifies the X designation. To be honest, it's a win-win. Intel will have a new wave of customers who will finally step up from the xx30 CPU to the X flagship, and in turn their profitability and profit margins will increase. Enthusiasts who desire to have the latest and greatest and also use the PC for productivity will like the move from 6 to 8 cores and will have the funds to pay for it.

Offering a 6-core 5820 is also big news for the desktop PC market since many people who use the PC for productivity (other than games) will now consider the X99 platform over Z97. For these users, Intel is also opening up an upgrade path to Broadwell-E - more profits for Intel and a significantly faster CPU for productivity for the consumer at the $400 and $1K levels.

If Intel released an 8-core at $500 level right away, it would cannibalize their 4770/4790K sales and reduce their profit margins. The only other possibility I saw was releasing an 8-core for $700-750. But I totally understand what they did. They are enticing some PC enthusiasts who care about productivity to choose the X99 platform with the 5820, are catering specifically to enthusiast gamers with 4790K, and also satisfying the ultra high end enthusiasts/productivity users with the 8-core as the flagship. It's a very smart strategy. Over time we'll get an 8-core for $500.

If Intel is really going to give us a ~$400 USD 6-core 5820, this will sell like hot cakes!

 

[SAAA]

If Intel released an 8-core at $500 level right away, it would cannibalize their 4770/4790K sales and reduce their profit margins.

I wonder sometimes how much that would actually damage their profit rather than increase it: think how many people would upgrade and pay 500-700$ instead of ~350$ for an 8 core desktop if that happened, propably enough to compensate the loss of not selling it at 1000$!
Of course they are doing something similar by selling a six core at probably 400$, the current price of 4820K is too competitive if upgraded to six-cores with any faster quad, they have the same architecture too...
Maybe with Broadwell we will see some more 8 and perhaps 10 cores -E CPUs.

 

[Shimmishim]

If anything what they did makes more sense. Before, they were charging $1K for more or less extra cache a small bump in clocks over the 4930. In terms of manufacturing, it doesn't cost them much more to make the 4960 vs. 4930, but how many enthusiasts had any incentive to buy the 4960X? Intel was losing out on a lot of profits. Now by moving the 8-core exclusively into the X realm this finally justifies the X designation. To be honest, it's a win-win. Intel will have a new wave of customers who will finally step up from the xx30 CPU to the X flagship, and in turn their profitability and profit margins will increase. Enthusiasts who desire to have the latest and greatest and also use the PC for productivity will like the move from 6 to 8 cores and will have the funds to pay for it.

Offering a 6-core 5820 is also big news for the desktop PC market since many people who use the PC for productivity (other than games) will now consider the X99 platform over Z97. For these users, Intel is also opening up an upgrade path to Broadwell-E - more profits for Intel and a significantly faster CPU for productivity for the consumer at the $400 and $1K levels.

If Intel released an 8-core at $500 level right away, it would cannibalize their 4770/4790K sales and reduce their profit margins. The only other possibility I saw was releasing an 8-core for $700-750. But I totally understand what they did. They are enticing some PC enthusiasts who care about productivity to choose the X99 platform with the 5820, are catering specifically to enthusiast gamers with 4790K, and also satisfying the ultra high end enthusiasts/productivity users with the 8-core as the flagship. It's a very smart strategy. Over time we'll get an 8-core for $500.

If Intel is really going to give us a ~$400 USD 6-core 5820, this will sell like hot cakes!

I completely agree that it does make sense but that doesn't mean I have to like it! I was really hoping that they'd make the X930 8-core but have less cache just like the current one 6-core CPUs. Now, I have to save up even more if I want to get the 8-core 5960X beast.

 

[SAAA]

That may not go as well as you'd think. I'm sure it will vary somewhat with each individual chip, but using my 3930K as an example, it didn't seem to have any spectacular or dud cores. I found this out while trying to push up some single threaded benchmarking scores. All six are stable @5.1, but even with the other cores disabled not a single one of them was stable @5.2 within the voltage range I was willing to push.

Wow, that chip is really something! Well at those frequencies you are hitting voltage limits more than thermal ones, but imagine for many worst cases were one can't even hit those speeds, if a single core could and if locked at a few bin higher would still be beneficial over a lower average clock. There's 8 chances with a 5960X to have a better core, hell if 4790K reaches 4.4GHz why spend 1000$ and don't even match that speed on a core at least?

 

[moonbogg]

I WANT to be excited about 4790k, but I can't help but suspect that quad cores are on their way out as the best gaming option. I think the performance delta between quad and hex is going to broaden very quickly with upcoming titles. Can't bring myself to go back to quad for this reason, regardless of how high it clocks. I'd like to be convinced since a 4790k would be a good priced option, but I think 6 cores is where its at for gaming pretty much.

 

[Rakehellion]

Give me an 8-core laptop.

 

[SAAA]

I WANT to be excited about 4790k, but I can't help but suspect that quad cores are on their way out as the best gaming option. I think the performance delta between quad and hex is going to broaden very quickly with upcoming titles. Can't bring myself to go back to quad for this reason, regardless of how high it clocks. I'd like to be convinced since a 4790k would be a good priced option, but I think 6 cores is where its at for gaming pretty much.

Hold tight that hex goodness, there's no need to change it until next tock after Skylake, if history teaches something the same happened with dual vs quads and with Nehalem and it's 6-cores.

Give me an 8-core laptop.

ARM or Atom ones?