Intel shows Core i7-6950X specs at 3,5GHz and 25MB cache

Published by

Click here to post a comment for Intel shows Core i7-6950X specs at 3,5GHz and 25MB cache on our message forum
https://forums.guru3d.com/data/avatars/m/255/255262.jpg
Business as usual then, if true. Intel is unable to actually make similar products better,and instead add a couple of cores and increase the price to an even more ridiculous point. Had the development and competition been healthy the 10 core processor would replace the old 8 core CPU at a similar price point. 1500$ is extremely much right now with the strong dollar and weak other currencies. Frustrating how all tech is priced in dollars, really.
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
I want More Cores for rendering..I don't want Xeons..I ran mine with 1.24v for 4.5ghz..:D
wow that's actually a very good chip :banana: my 4.5 still linger at around 1.28-1.29, currently running at 1.3 which wasn't totally stable at 4.6 😉
data/avatar/default/avatar27.webp
I'm definitely going to get one of these... .... when I win the lottery 🙂
https://forums.guru3d.com/data/avatars/m/262/262613.jpg
wow that's actually a very good chip :banana: my 4.5 still linger at around 1.28-1.29, currently running at 1.3 which wasn't totally stable at 4.6 😉
and yours is the 6 core variant at 1.3, i'm a little skeptical though about the 8 core @ 1.24V. I'd have to see it for my self.
https://forums.guru3d.com/data/avatars/m/164/164785.jpg
and yours is the 6 core variant at 1.3, i'm a little skeptical though about the 8 core @ 1.24V. I'd have to see it for my self.
Yah 4.6 @ 1.24V on the 5960X is BS, even a super binned chip won't yield those results.
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
I'm quite lucky and I can get 1.2v to run at 4.5, for some reason needs 1.225v if I up cache to 4GHz, even if I don't change uncore voltage, it just isn't stable with core 4.5 at 1.2vcore [SPOILER]http://i.imgur.com/baLeDuT.png[/SPOILER]
https://forums.guru3d.com/data/avatars/m/56/56686.jpg
Intel Core i7-6800K I would like to know when this is coming out and if it to replace the 5820k? if it less then the 6700k I might get that instead, I do hope it able to sqeeze out an extra 100 STP
data/avatar/default/avatar33.webp
Will they finally call it an i9?
Now this I think is a good idea! 10/core/20 Threads emmmmm, think of all the diffrent (porn) videos I can watch at the same time with this chip.I am getting one, there is nothing else to talk about lol.:wanker:
https://forums.guru3d.com/data/avatars/m/259/259654.jpg
yup, if zen goes for ipc and outdoes the top i3 or lower i5, they'll be back in business. The usage trend and the user requirement of that market doesnt even need hyper threading, just IPC will hold up, as has intel's ! Well must be a month or two away i guess
The killing move would be some HBM on board. That would increase the processor speed tremendously (especially in gaming, see the 128MB Broadwell that was getting 4790k speeds), and with a similar IPC to Haswell/Ivy and a good price, then maybe FINALLY the market would normalize. These products now are sold like this because Intel can sell them like this. I'm impressed they didn't try to gouge out another socket.
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
The killing move would be some HBM on board. That would increase the processor speed tremendously (especially in gaming, see the 128MB Broadwell that was getting 4790k speeds), and with a similar IPC to Haswell/Ivy and a good price, then maybe FINALLY the market would normalize. These products now are sold like this because Intel can sell them like this. I'm impressed they didn't try to gouge out another socket.
AMD claims 40% ipc, if we assume it's accurate that puts it 20% slower than haswell going by single thread benchmarks
https://forums.guru3d.com/data/avatars/m/259/259654.jpg
AMD claims 40% ipc, if we assume it's accurate that puts it 20% slower than haswell going by single thread benchmarks
That 40% actually puts it on par with Haswell. It's supposed to be 40% IPC over Excavator, which is the 4th iteration of Bulldozer. I can't wait until we have something more concrete. 20% slower than Haswell might be enough, depending on cache/onboard memory, multithreading performance and overclockability.
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
I can't wait until we have something more concrete. 20% slower than Haswell might be enough, depending on cache/onboard memory, multithreading performance and overclockability.
Yeah if they can sell an eight core with SMT at 80% ipc as haswell for 400$ they're golden, 4ghz I would hope
https://forums.guru3d.com/data/avatars/m/259/259654.jpg
Yeah if they can sell an eight core with SMT at 80% ipc as haswell for 400$ they're golden
That 40% actually puts it on par with Haswell. It's supposed to be 40% IPC over Excavator, which is the 4th iteration of Bulldozer. My inner geek still wants to see some HBM on that thing. If it has it, I'm sure it will destroy even Skylake if the non-HBM IPC numbers are like this. They are less afraid than Intel to have a 150W desktop CPU, so it might actually get higher than 4GHz (remember it's 14nm). http://i.imgur.com/2bFtWbn.png
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
I'm assuming the fx9590 is excavator? By my calculations 40% extra ipc put it below haswell, did you account for clock differences? Nope it isn't, do you have single thread benches for an excavator part? Thats a nice diagram!
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
and yours is the 6 core variant at 1.3, i'm a little skeptical though about the 8 core @ 1.24V. I'd have to see it for my self.
Not sure if he posted screens of his overclocks, but I remember him getting a 5960X after he was so disappointed with his 5930K that couldn't even reach 4.5 iirc.
I'm quite lucky and I can get 1.2v to run at 4.5, for some reason needs 1.225v if I up cache to 4GHz, even if I don't change uncore voltage, it just isn't stable with core 4.5 at 1.2vcore [SPOILER]http://i.imgur.com/baLeDuT.png[/SPOILER]
Nice one there!
data/avatar/default/avatar27.webp
Here I am wondering if Intel ever plans to release/create anything making a replacement of an old 2600K/3570K a worthwhile endeavor. The wait continues 😉
https://forums.guru3d.com/data/avatars/m/142/142454.jpg
Here I am wondering if Intel ever plans to release/create anything making a replacement of an old 2600K/3570K a worthwhile endeavor. The wait continues 😉
It depends on your use and setup but I as well as others got significant improvements going from a SandyBridge to Haswell/Skylake.
data/avatar/default/avatar23.webp
That 40% actually puts it on par with Haswell. It's supposed to be 40% IPC over Excavator, which is the 4th iteration of Bulldozer. My inner geek still wants to see some HBM on that thing. If it has it, I'm sure it will destroy even Skylake if the non-HBM IPC numbers are like this. They are less afraid than Intel to have a 150W desktop CPU, so it might actually get higher than 4GHz (remember it's 14nm). http://i.imgur.com/2bFtWbn.png
I'll wait for actual user's numbers, recall AMD claiming bulldozer was x amount better than phenom back in the day.
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
I'm assuming the fx9590 is excavator? By my calculations 40% extra ipc put it below haswell, did you account for clock differences? Nope it isn't, do you have single thread benches for an excavator part? Thats a nice diagram!
There are no ST or any other reliable benchmark scores for Excavator except for Athlon X4 845 as others came in form of TDP limited mobile chips. But I remember thread of person on OCN who had development board, then unlocked TDP and it gained around ~5% in comparison to what he had before. (But that was mobile chip in desktop MB with proper cooling to begin with.) So, I see it is not that easy to guesstimate ST performance + IPC is not everything there is to CPU... And on top of that Zen 'drops' clusters which may have effect on ST performance + new cache system. And while Lisa Su in some video mentioned that it came out better than expected, AMD actually does not need to match or outperform intel's offering in every benchmark. They just need to outperform Sandy by 10~20% and deliver increased number of cores at reasonable price. AMD gave us only one 'desktop' Carrizo. Athlon X4 845 (locked) 65W only 8 PCIe 3.0 lanes, and yes that's on 28nm. http://browser.primatelabs.com/geekbench3/compare/5624302?baseline=2618832 On left nearly best result it did (likely somewhere between 4~4.2GHz). On right my system @4.5GHz. Then: http://hwbot.org/hardware/processor/athlon_x4_845/ Here, some MT (clusters) results and clock people could squeeze out of locked chip. https://www.youtube.com/watch?v=vlPSz2pD-zw&&t=5m25s Anand has thread here: http://forums.anandtech.com/showthread.php?t=2465958 at 3.8GHz it has CB11.5 => 1.07 || my i5 @4.5GHz does 1.74 (Clock adjustment 1.07 *4.5/3.8 = 1.27 || +40% => 1.77) And Cinebench is actually favoring intel CPUs as its score intel:AMD ratio is always looking much better for intel than any other benchmark.