Intel 7900x and Gigabyte X299 AORUS Gaming 7 leaked in benchmark

[done12many2]

Dup

[PCGuy_5960]

5 minutes ago, done12many2 said:

I'd argue that the 5960x is equally as good for gaming as the 5820k if not better.  

 

Binning has a lot to do with the decent OC that you mentioned and 5960x CPUs are the top binned Haswell-E.

I think he meant that the 5960X isn't a sensible choice for gaming, but the 5820K is. The 5960X is of course better, but you're paying 2-3x as much for slightly better performance.

6 minutes ago, done12many2 said:

One of the best kept secrets with regards to gaming chips is the Broadwell 5775c/5675c with their ridiculous 128MB of eDRAM, which effectively serves as a massive cache.  Those chips at low clock speeds easily outpaced Skylake chips at much higher speeds with the cache alone.  

Interesting! 

[PCGuy_5960]

10 minutes ago, Valentyn said:

One can hope, because at that price it'll finally be worth it. Just a shame about the 28 PCIe lanes. 

28 CPU lanes+24 chipset lanes  So 52 lanes in total

[79wjd]

15 minutes ago, done12many2 said:

 

 One of the best kept secrets with regards to gaming chips is the Broadwell 5775c/5675c with their ridiculous 128MB of eDRAM, which effectively serves as a massive cache.  Those chips at low clock speeds easily outpaced Skylake chips at much higher speeds with the cache alone.  

I thought Broadwell performed comparatively poorly compared to Haswell/Skylake i7's because they have the cache layout of an i5 (eDRAM excluded).

[Coaxialgamer]

16 hours ago, done12many2 said:

I think it's funny that anyone would think that power savings is high on the priority list for folks who use HEDT.  

That power still needs to be exhausted as heat though...

[done12many2]

10 minutes ago, djdwosk97 said:

I thought Broadwell performed comparatively poorly compared to Haswell/Skylake i7's because they have the cache layout of an i5 (eDRAM excluded).

 

That little sucker held it's own especially when you compare it clock for clock.  Like I said, it was highly overlook by most for gaming purposes.  Not saying it was the best thing out there, but damn good.  Overclock it a bit and it makes for a badass little gaming CPU.

 

http://www.anandtech.com/show/9320/intel-broadwell-review-i7-5775c-i5-5675c/8

 

http://www.pcworld.com/article/2950704/hardware/intel-core-i7-5775c-review-the-unwanted-desktop-broadwell-has-one-neat-trick.html?page=3

 

[done12many2]

9 minutes ago, Coaxialgamer said:

That power still needs to be exhausted as heat though...

 

Same thing applies with both brands.  If you'll read back a few posts, you'll note that in loaded conditions at equal clockspeeds, the amount of power consumed is the same.  You should look at the context of my post as it relates to those posts before it.  

[79wjd]

2 minutes ago, done12many2 said:

 

That little sucker held it's own especially when you compare it clock for clock.  Like I said, it was highly overlook by most for gaming purposes.  Not saying it was the best thing out there, but damn good.  Overclock it a bit and it makes for a badass little gaming CPU.

 

http://www.anandtech.com/show/9320/intel-broadwell-review-i7-5775c-i5-5675c/8

 

http://www.pcworld.com/article/2950704/hardware/intel-core-i7-5775c-review-the-unwanted-desktop-broadwell-has-one-neat-trick.html?page=3

 

PCWorld was looking at iGPU performance, so of course Broadwell would be great there and Anandtech tested with a 770/285....for some reason even though the Titan XM was out for months. 

[done12many2]

1 minute ago, djdwosk97 said:

PCWorld was looking at iGPU performance, so of course Broadwell would be great there and Anandtech tested with a 770/285....for some reason even though the Titan XM was out for months. 

 

I think you're misunderstanding my point, which is nothing more then it was overlooked as a viable gaming CPU.  Clocks speed were low on it, but the tons of eDRAM still helped it with discrete GPUs as well.  

 

 

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/10

 

http://www.anandtech.com/show/10968/the-intel-core-i7-7700k-91w-review-the-new-stock-performance-champion/6

 

https://www.purepc.pl/procesory/broadwell_niszczyciel_test_core_i5_5675c_i_core_i7_5775c?page=0,38

 

 

[Senzelian]

4 hours ago, Drak3 said:

False. It has quad IMCs. Each CCX uses a single IMC, with Ryzen having 2 total, allowing it dual channel RAM.

When did AMD say that each CCX has its own IMC?

[Drak3]

2 hours ago, Senzelian said:

When did AMD say that each CCX has its own IMC?

They didn't, but the die shot actually shows 2 separate IMCs, one in each CCX.

[Drak3]

8 hours ago, PCGuy_5960 said:

I think he meant that the 5960X isn't a sensible choice for gaming, but the 5820K is. The 5960X is of course better, but you're paying 2-3x as much for slightly better performance.

@done12many2 Partially correct. In relation to the 5820K, it achieves nearly 70MHz higher OCs at average on water, and 100MHz higher on air, but it's ~3x the price, and the extra cores currently don't yield much benefit much in gaming alone, as a matter of fact, depending on the scheduler, having the extra cores might take a percentage or 2 off of performance. If one were to make a gaming oriented KVM, I'd say that the 5960X is a good choice, but for straight gaming, it's not that good.

Then again, the 5930K's averages are 4.44GHz air, and 4.78GHz water, and it has the same 40 PCIe lanes of the 5960X without being a bank breaker.

 

Only reason I can EASILY recommend the R7 1700 or 1800X for a mainly gaming build is that they'll deliver avg. and min. framerates high enough for a damn good gaming experience, great frame times, they cost around $340-360 and $500 respectively, and the Hexacore/Quadcore w/ SMT don't clock better. While the frame rates and times are similar on Intel HEDT, both price brackets are occupied by Hexacores.

8 hours ago, done12many2 said:

One of the best kept secrets with regards to gaming chips is the Broadwell 5775c/5675c with their ridiculous 128MB of eDRAM, which effectively serves as a massive cache.  Those chips at low clock speeds easily outpaced Skylake chips at much higher speeds with the cache alone

It really depends on the task though. In some tasks, it was practically a worse clocking Haswell with a good iGPU.

 

8 hours ago, done12many2 said:

It those two factors play a role in Skylake X, one can expect some nice performance especially when combined with the advantages of quad channel memory and quadrupled L2 cache.

I still don't expect Intel to be able to push a 4GHz Decacore or Dodecacore without a TDP runaway, and it might take an MCM to better spread the heat of the chip. Not to mention power draw. I kind of expect that a 16C ThreadRipper and a 12C Skylake-X being almost on par, slight advantage Skylake, at multicore performance. Should that prove true, Intel has the single core and quad core performance advantages, which will lend credibility to them for more situations. ThreadRipper would be more practical if single core/quad core performance takes a back seat to cores.

[done12many2]

7 minutes ago, Drak3 said:

@done12many2 Partially correct. In relation to the 5820K, it achieves nearly 70MHz higher OCs at average on water, and 100MHz higher on air, but it's ~3x the price, and the extra cores currently don't yield much benefit much in gaming alone, as a matter of fact, depending on the scheduler, having the extra cores might take a percentage or 2 off of performance. If one were to make a gaming oriented KVM, I'd say that the 5960X is a good choice, but for straight gaming, it's not that good.

Then again, the 5930K's averages are 4.44GHz air, and 4.78GHz water, and it has the same 40 PCIe lanes of the 5960X without being a bank breaker.

 

I find your "average" numbers interesting.  Out of the 2 5960x CPUs I've had, I've rarely found 5820k or 5930k owners able to match my clockspeeds.  There are a couple, but beyond rare.  With that said, clock for clock and core for core, they would perform the same if not potentially better regardless of how many extra core a 5960x has.  When I was addressing your statement, I did not factor in value as you didn't originally mention that.   

 

Where did you pull your statistics from?

 

Here's the leaderboard from OCN.

 

http://www.overclock.net/t/1510388/haswell-e-overclock-leaderboard-owners-club/0_100

 

7 minutes ago, Drak3 said:

Only reason I can EASILY recommend the R7 1700 or 1800X for a mainly gaming build is that they'll deliver avg. and min. framerates high enough for a damn good gaming experience, great frame times, they cost around $340-360 and $500 respectively, and the Hexacore/Quadcore w/ SMT don't clock better. While the frame rates and times are similar on Intel HEDT, both price brackets are occupied by Hexacores.

 

Can't argue with that at all.  Best bang for the buck there is for an overall well-balanced chip.

 

7 minutes ago, Drak3 said:

It really depends on the task though. In some tasks, it was practically a worse clocking Haswell with a good iGPU.

 

"it depends" applies to just about everything in life.  My point was that it was a great little gaming chip at the time that didn't get much attention.

 

7 minutes ago, Drak3 said:

I still don't expect Intel to be able to push a 4GHz Decacore or Dodecacore without a TDP runaway, and it might take an MCM to better spread the heat of the chip. Not to mention power draw. I kind of expect that a 16C ThreadRipper and a 12C Skylake-X being almost on par, slight advantage Skylake, at multicore performance. Should that prove true, Intel has the single core and quad core performance advantages, which will lend credibility to them for more situations. ThreadRipper would be more practical if single core/quad core performance takes a back seat to cores.

 

I think you and I are approaching this from two different directions, which is no big deal.  The only think I really care about it performance at max overclock.  I don't expect Intel to come out of the box with a 4.5 GHz 10 and 12 core part.  What I do expect is that based on the new architecture, overclocking will be at least in keeping with current expectations of OC headroom if not better.  

 

Most Intel Extreme parts are very well known for OC headroom far greater then 1 GHz over stock clocks.  If this continues, and you combine the rest of the stuff I mentioned, we're in for some fun ahead.  

 

[Drak3]

1 minute ago, done12many2 said:

Where did you pull your statistics from?

HWBot. The only issue I have with it is that the 5930K's average on air and water are flipped.

http://hwbot.org/hardware/processor/core_i7_5930k/

http://hwbot.org/hardware/processor/core_i7_5960x/

4 minutes ago, done12many2 said:

With that said, clock for clock and core for core, they would perform the same if not potentially better regardless of how many extra core a 5960x has.  When I was addressing your statement, I did not factor in value as you didn't originally mention that.

If the scheduler isn't the spawn of a retard and a sadist, then theoretically it won't matter. Also, ignoring value, I don't consider octocores to be the optimal gaming CPUs yet. Keyword, yet. We're heading there, if slowly.

8 minutes ago, done12many2 said:

The only think I really care about it performance at max overclock.  I don't expect Intel to come out of the box with a 4.5 GHz 10 and 12 core part.  What I do expect is that based on the new architecture, overclocking will be at least in keeping with current expectations of OC headroom if not better.

I don't mean stock. I cannot see a CPU having 3x the cores of its quadcore borther, more memory channels, more PCIe lanes, AND maintain the same clocks at the same voltages without overheating. I can see a Dodecacore hitting 4.5-4.8GHz on cascade cooling, but not on a water loop.

 

10 minutes ago, done12many2 said:

Most Intel Extreme parts are very well known for OC headroom far greater then 1 GHz over stock clocks.  If this continues, and you combine the rest of the stuff I mentioned, we're in for some fun ahead.

Yes, but how many HEDT parts are shipping with overly high clocks? Sure, I have a 1GHz offset on my 5930K, and can go further if I up my cooling, but it comes as a 3.5GHz part. The highest clocked X99 part is only 3.6GHz

[done12many2]

1 hour ago, Drak3 said:

HWBot. The only issue I have with it is that the 5930K's average on air and water are flipped.

http://hwbot.org/hardware/processor/core_i7_5930k/

http://hwbot.org/hardware/processor/core_i7_5960x/

 

It's honestly hard to do this when you mention one CPU in one post, but provide stats for another CPU in another post.  You were originally talking about a 5820k and a 5960x.

 

1 hour ago, Drak3 said:

@done12many2 Partially correct. In relation to the 5820K, it achieves nearly 70MHz higher OCs at average on water, and 100MHz higher on air, but it's ~3x the price, and the extra cores currently don't yield much benefit much in gaming alone, as a matter of fact, depending on the scheduler, having the extra cores might take a percentage or 2 off of performance. If one were to make a gaming oriented KVM, I'd say that the 5960X is a good choice, but for straight gaming, it's not that good.

Then again, the 5930K's averages are 4.44GHz air, and 4.78GHz water, and it has the same 40 PCIe lanes of the 5960X without being a bank breaker.

 

Using your own sources, you were incorrect in 100 MHz higher on air and 70 MHz higher on water stats.

 

Per your sources, the 5820k is on average 254 107 MHz slower then a 5960x when both are cooled with air and 65 MHz slower when both are cooled with water.  

 

You were also incorrect about your stats on the 5930k.  Per your sources again, the 5930k averages 4.44 GHz on water and 4.78 GHz on air.  Not the other way around as you stated.  It is indeed that air is 300+ MHz faster on average then water.  Do you see any problems with your source?

 

That's why I linked you to the leaderboard from OCN.

 

This entire thing stemmed from the original statement below.  Since then, you've added value and other considerations that you did not originally mention in your first comment about the 5960x as a gaming CPU.  Taking this down to the very basics, my response to your original statement was appropriate and accurate. 

 

If your first post would have mentioned value and the other stuff that you later brought up, I would have never commented on it to begin with.  After all, within the context of @PCGuy_5960's post, which sparked your original comment about the 5960x, value was not a talking point.  He asked "Who buys HEDT CPUs and only plays games?"  

 

I ask, "Who the hell buys HEDT for value?"

 

14 hours ago, PCGuy_5960 said:

Who buys HEDT CPUs and only plays games? 

11 hours ago, Drak3 said:

Depends on the HEDT part. I wager that the 5820K, with a decent OC, is a great option for gaming, moving forward. I wouldn't say that about the 5960X.

10 hours ago, done12many2 said:

I'd argue that the 5960x is equally as good for gaming as the 5820k if not better.    

 

It really didn't need to get this complicated, but you and I have been down similar roads before.  I'll just shrug my shoulders and move on.  

 

5820k air:

 

 

5960x air:

 

 

5820 water:

 

 

5960x water:

 

 

 

5930k stats:

 

 

[dexT]

13 minutes ago, done12many2 said:

 

It's honestly hard to do this when you mention one CPU in one post, but provide stats for another CPU in another post.  You were originally talking about a 5820k and a 5960x.

 

 

Using your own sources, you were incorrect in 100 MHz higher on air and 70 MHz higher on water stats.

 

Per your sources, the 5820k is on average 254 107 MHz slower then a 5960x when both are cooled with air and 65 MHz slower when both are cooled with water.  

 

You were also incorrect about your stats on the 5930k.  Per your sources again, the 5930k averages 4.44 GHz on water and 4.78 GHz on air.  Not the other way around as you stated.  It is indeed that air is 300+ MHz faster on average then water.  Do you see any problems with your source?

 

That's why I linked you to the leaderboard from OCN.

 

This entire thing stemmed from the original statement below.  Since then, you've added value and other considerations that you did not originally mention in your first comment about the 5960x as a gaming CPU.  Taking this down to the very basics, my response to your original statement was appropriate and accurate. 

 

If your first post would have mentioned value and the other stuff that you later brought up, I would have never commented on it to begin with.  After all, within the context of @PCGuy_5960's post, which sparked your original comment about the 5960x, value was not a talking point.  He asked "Who buys HEDT CPUs and only plays games?"  

 

I ask, "Who the hell buys HEDT for value?"

 

 

It really didn't need to get this complicated, but you and I have been down similar roads before.  I'll just shrug my shoulders and move on.  

 

5820k air:

 

 

5960x air:

 

 

5820 water:

 

 

5960x water:

 

 

 

5930k stats:

 

 

Come at me brosicle. Pifast 6 vs 8 Core

http://hwbot.org/submission/3522108_

[done12many2]

Just now, dexT said:

Come at me brosicle. Pifast 6 vs 8 Core

http://hwbot.org/submission/3522108_

 

I like your style buddy!  DICE wins.

 

Do you have that bad boy loaded now on ambient?  

[dexT]

Just now, done12many2 said:

 

I like your style buddy!  DICE wins.

 

Do you have that bad boy loaded now on ambient?  

Naw I traded my 5820K for a X48 REX plus 20+ cpus and 4 sticks of binned golden DDR2. He also threw in 5 tubes IC Diamond.

[done12many2]

Just now, dexT said:

Naw I traded my 5820K for a X48 REX plus 20+ cpus and 4 sticks of binned golden DDR2. He also threw in 5 tubes IC Diamond.

 

Nice.

 

You won't want my 5960x then.  It's garbage and overclocks low compared to your 5820k.  

[dexT]

Just now, done12many2 said:

 

Nice.

 

You won't want my 5960x then.  It's garbage and overclocks low compared to your 5820k.  

Naw I'll rock that. Once it's sub 5 hundy though. I have some VGA dat needs teh cores.

[done12many2]

9 minutes ago, dexT said:

Naw I'll rock that. Once it's sub 5 hundy though. I have some VGA dat needs teh cores.

 

I just took interest in folding not too long ago.  I've been doing a little for LTT and quite a bit for OCN.  

 

I usually keep two faster rigs on hand besides the other household stuff.  Right now that's my x99 / 5960x and my z270 / 7700k.  I've been thinking about a Skylake X and a Threadripper combo for the next round.  Since the x299 platform will support a higher TDP 4c and bigger chips, I won't need the z270 platform anymore.  

 

If that's what ends up happening, I'll more the likely work a deal out with you bud.  I can't tell you how it would do on cold, but I can tell you that on ambient it's an absolute beast.  

[dexT]

Just now, done12many2 said:

 

I just took interest in folding not too long ago.  I've been doing a little for LTT and quite a bit for OCN.  

 

I'm usually keep two faster rigs on hand besides the other household stuff.  Right now that's my x99 / 5960x and my z270 / 7700k.  I've been thinking about a Skylake X and Threadripper combo for the next round.  Since the x299 platform will support a higher TDP 4c and bigger chips, I won't need the z270 platform anymore.  

 

If that's what ends up happening, I'll more the likely work a deal out with you bud.  I can't tell you how it would do on cold, but I can tell you that on ambient it's an absolute beast.  

Hey it has major potential. Sometimes an ambiant god is garbage sub-zero though but I'm willing to chance

[dexT]

Just now, done12many2 said:

 

I just took interest in folding not too long ago.  I've been doing a little for LTT and quite a bit for OCN.  

 

I'm usually keep two faster rigs on hand besides the other household stuff.  Right now that's my x99 / 5960x and my z270 / 7700k.  I've been thinking about a Skylake X and Threadripper combo for the next round.  Since the x299 platform will support a higher TDP 4c and bigger chips, I won't need the z270 platform anymore.  

 

If that's what ends up happening, I'll more the likely work a deal out with you bud.  I can't tell you how it would do on cold, but I can tell you that on ambient it's an absolute beast.  

I'm dreaming of X299 and Threadripper myself.

[done12many2]

7 minutes ago, dexT said:

Hey it has major potential. Sometimes an ambiant god is garbage sub-zero though but I'm willing to chance

 

Do you have a z170 or z270 board?

[Drak3]

40 minutes ago, done12many2 said:

Per your sources, the 5820k is on average 254 107 MHz slower then a 5960x when both are cooled with air and 65 MHz slower when both are cooled with water.

Unless you're saying that the 5820K is an octocore going for ~1000 USD, and the 5960X goes for ~360USD with only 6 cores, I'd read my statement again, carefully.

43 minutes ago, done12many2 said:

You were also incorrect about your stats on the 5930k.  Per your sources again, the 5930k averages 4.44 GHz on water and 4.78 GHz on air.  Not the other way around as you stated.  It is indeed that air is 300+ MHz faster on average then water.  Do you see any problems with your source?

The problem with my source is that, for the 5930K, the values are mislabeled, which I said occured with that particular CPU on HWBot. It should read that the average for air is 4.44GHz and that the average for water if 4.78. However, if one actually takes a moment, and is familiar with the subject at hand, they can figure out a simple error of mislabeling one thing.

 

49 minutes ago, done12many2 said:

This entire thing stemmed from the original statement below.  Since then, you've added value and other considerations that you did not originally mention in your first comment about the 5960x as a gaming CPU.

My initial post was not a full 'considerations given' statement. It was a very brief responce as to why someone would use an HEDT CPU for gaming. I responded with a specific sku, with a decent overclock, can be a good option, and a different, specific sku isn't a good option in my opinion. I didn't actually specify why (albeit if anyone remembers my more recent ramblings, they'll remember why), because I wanted my responce to just be a short snippet. Someone wants an elaboration on my opinion, and I'll give them one.

 

When it came down to a difference of less than 100MHz on water, I introduced the third sku that averages better than both, because apparently, it was coming down to clockspeeds as being the big factor.

 

53 minutes ago, done12many2 said:

It's honestly hard to do this when you mention one CPU in one post, but provide stats for another CPU in another post.  You were originally talking about a 5820k and a 5960x.

Yes, but here's the thing. I gave the link to the 5930K to show you the only issue I have with HWBot's stats. A clerical error. And, I pointed out, before giving the link, that I had issue with said error, for what should be obvious reason.

Otherwise, I'd just have given you the link to the 5960X. I know that you know a good deal of the finer points of computing. If you can't navigate the site, there's not much I can do, and I'm not the type to spoon feed everyone everything.

 

1 hour ago, done12many2 said:

Who the hell buys HEDT for value?

That depends on what you value. When I bought my 5930K, I valued having a system with more power than I needed at the time, but also had 2 extra cores to leverage CPU affinity, as well as maybe start game streaming or get back into animating. I valued having a system that was fun to build, and could easily be adapted into a workstation.