Razer Leviathan V2 gaming soundbar review
Guru3D NVMe Thermal Test - the heatsink vs. performance
EnGenius ECW220S 2x2 Cloud Access Point review
Alphacool Eisbaer Aurora HPE 360 LCS cooler review
Noctua NH-D12L CPU Cooler Review
Silicon Power XPOWER XS70 1TB NVMe SSD Review
Hyte Y60 chassis review
ASUS ROG Thor 1000W Platinum II (1000W PSU) review
ASUS ROG Rapture GT-AXE11000 WIFI6E router review
Backforce One Plus Gaming Chair review
Socket 2066 (Core i9-7960X) Processor with 16 cores surfaces in GeekBench
In the GeekBench ranking list a new 16-core processor with 32-threads has surfaced. The unit is tagged as Socket 2066 and thus is the recently announced (yet not available) Core i9-7960X. The results reveal a few interesting things.
The $1699 Core i9-7960X Skylake-E processor is identified as having 16 cores and 32 threads. The base clock frequency is 2.51 GHz, indicative for an engineering sample and perhaps with this many cores, it might even be the final base-clock. The chip has 1MB L2-cache per core and 22 MB of shared L3 cache in-between the cores as you can see in the screenshot.
The twp scores are as follows:
- Single score results: 5238 points
- Multi-threaded score: 33672 points
So when you compare a little back and forth with the ranking lists then that single threaded score brings the processor at the level of a i5-7600 perf wise, and the multi-threaded score actually brings it close to the 10-core Core i9 7900X. That processor obviously has a base clock of 3.3 GHz. With a little Turbo tweaking we do expect the final performance to be better though.
| Clock | Turbo 2.0/3.0 | Cores / threads | L3-cache | PCIe 3.0 | Mem Channels | TDP | Price | |
|---|---|---|---|---|---|---|---|---|
| Core i9 7980XE | TBA | TBA | 18/36 | TBA | TBA | Quad-channel DDR4-2666 | TBA | $1999 |
| Core i9 7960X | 2.5 GHz | TBA | 16/32 | 22MB | TBA | Quad-channel DDR4-2666 | TBA | $1699 |
| Core i9 7940X | TBA | TBA | 14/28 | TBA | TBA | Quad-channel DDR4-2666 | TBA | $1399 |
| Core i9 7920X | TBA. | TBA | 12/24 | TBA | TBA | Quad-channel DDR4-2666 | TBA | $1199 |
| Core i9 7900X | 3.3 GHz | 4.3/4.5 GHz | 10/20 | 13,75 MB | 44 | Quad-channel DDR4-2666 | 140 W | $999 |
| Core i7 7820X | 3.6 GHz | 4.3/4.5 GHz | 8/16 | 11 MB | 28 | Quad-channel DDR4-2666 | 140 W | $599 |
| Core i7 7800X | 3.5 GHz | 4.0 GHz | 6/12 | 8,25 MB | 28 | Quad-channel DDR4-2666 | 140 W | $389 |
| Core i7 7740X | 4.3 GHz | 4.5 GHz | 4/8 | 8 MB | 16 | Dual-channel DDR4-2666 | 112 W | $339 |
| Core i5 7640X | 4.0 GHz | 4.2 GHz | 4/4 | 6 MB | 16 | Dual-channel DDR4-2666 | 112 W | $242 |
ASUS's Dual Socket 2011 Workstation - 02/24/2012 01:09 PM
With the pending launch of the Sandy Bridge-EP Xeon E5 platform,
ASUS shows Z9PE-D8 WS dual-socket 2011 mobo - 01/10/2012 01:25 PM
Performance zealots will remember the ASUS Z7S WS which went head on with then Intel 'Skulltrail' D5400XS Xeon luxury gaming platform (using dual unlocked QX9775 chips), but since then ASUS has done n...
Veteran
Senior Member
Posts: 12103
Joined: 2009-01-16
Senior Member
Posts: 12103
Joined: 2009-01-16
#5455065 Posted on: 07/24/2017 05:40 PM
Funny - my i7-4930K ( 6/12 core ) / LGA2011 / X79 system running at a moderate 4.2Ghz overclock and built in 2014 produces a geekbench score of 4271 / 22463 ( single / multi )
cpu did cost me €500 back then...
Mines running 5ghz 24/7 and completely stable. Will not be changing for a good while yet
Funny - my i7-4930K ( 6/12 core ) / LGA2011 / X79 system running at a moderate 4.2Ghz overclock and built in 2014 produces a geekbench score of 4271 / 22463 ( single / multi )
cpu did cost me €500 back then...
Mines running 5ghz 24/7 and completely stable. Will not be changing for a good while yet
D3M1G0D
Senior Member
Posts: 2068
Joined: 2017-03-10
Senior Member
Posts: 2068
Joined: 2017-03-10
#5455066 Posted on: 07/24/2017 05:41 PM
Because its supposed to be reliable and safe to run at stock speeds. The CPU is overclockable so the stock speeds are kind of irrelevant anyway.
Honestly, I think the only reason Intel uses such high stock speeds for products like the 7700K is to make reviews look better. Many reviewers do a lot of their testing via stock speeds (where OC results may only have a page or 2 of content) so the only way for Intel to make each generation look better is to maintain a high base clock.
What I personally don't understand is why Intel doesn't just ditch base clocks entirely and have the turbo speed reach some obscenely high value (in terms of from-factory specs) like 5GHz. Turbo Boost at this point is designed to adjust for your power delivery, cooling, and workload. So why have a base clock anymore? Why not just have the CPUs run as fast as reliably possible? If power consumption is your concern, just use a weaker heatsink - problem solved. Or instead, maybe the motherboard can detect if you're using a 4-pin or 8-pin CPU power connector, where it will limit the clocks based on which one you're using. Against AMD, this could be a serious win for Intel since there's no way to directly compare the CPUs (without handicapping the Intel), and AMD doesn't have a CPU that can clock so high. I guess the main downside to this idea is Intel has been doing such a crappy job at improving IPC lately, so if they decided to go this route every generation, people would be even less convinced to upgrade.
The base clock is what the CPU is rated at for its TDP. It is the baseline specs, which guarantees a certain level of performance, and needs to be specified somewhere. Don't forget that OC is dependent on the CPU (silicon lottery) and the cooling, and Intel cannot sell their products based on what's theoretically possible from their silicon and with third-party coolers and such.
Overclocking is running a CPU beyond spec, and Intel takes no responsibility for it, and for good reason - to do what you are suggesting would mean Intel would be responsible for any damages that happen and would be liable (the lawsuits would be unending!). It would be an absolute disaster for Intel. Also, most people do not overclock, and want reliable and consistent performance for their products - like any product, there needs to be a guarantee in order for consumers to have confidence in a product.
Because its supposed to be reliable and safe to run at stock speeds. The CPU is overclockable so the stock speeds are kind of irrelevant anyway.
Honestly, I think the only reason Intel uses such high stock speeds for products like the 7700K is to make reviews look better. Many reviewers do a lot of their testing via stock speeds (where OC results may only have a page or 2 of content) so the only way for Intel to make each generation look better is to maintain a high base clock.
What I personally don't understand is why Intel doesn't just ditch base clocks entirely and have the turbo speed reach some obscenely high value (in terms of from-factory specs) like 5GHz. Turbo Boost at this point is designed to adjust for your power delivery, cooling, and workload. So why have a base clock anymore? Why not just have the CPUs run as fast as reliably possible? If power consumption is your concern, just use a weaker heatsink - problem solved. Or instead, maybe the motherboard can detect if you're using a 4-pin or 8-pin CPU power connector, where it will limit the clocks based on which one you're using. Against AMD, this could be a serious win for Intel since there's no way to directly compare the CPUs (without handicapping the Intel), and AMD doesn't have a CPU that can clock so high. I guess the main downside to this idea is Intel has been doing such a crappy job at improving IPC lately, so if they decided to go this route every generation, people would be even less convinced to upgrade.
The base clock is what the CPU is rated at for its TDP. It is the baseline specs, which guarantees a certain level of performance, and needs to be specified somewhere. Don't forget that OC is dependent on the CPU (silicon lottery) and the cooling, and Intel cannot sell their products based on what's theoretically possible from their silicon and with third-party coolers and such.
Overclocking is running a CPU beyond spec, and Intel takes no responsibility for it, and for good reason - to do what you are suggesting would mean Intel would be responsible for any damages that happen and would be liable (the lawsuits would be unending!). It would be an absolute disaster for Intel. Also, most people do not overclock, and want reliable and consistent performance for their products - like any product, there needs to be a guarantee in order for consumers to have confidence in a product.
schmidtbag
Senior Member
Posts: 6555
Joined: 2012-11-10
Senior Member
Posts: 6555
Joined: 2012-11-10
#5455073 Posted on: 07/24/2017 05:54 PM
I understand that, but if my idea were to be done then consumers would basically just pick the heatsink and/or PSU that best suits the TDP they're willing to have. Like I said, this would be based on a clock that can be reliably achieved, so basically Intel would take the worst-case scenario of a CPU on liquid cooling, drop the frequency 100Mhz or so, and that would be the maximum speed of the CPU (without overclocking). Silicon lottery is becoming less and less relevant for the average user. I get the impression most 7700Ks can reach 5GHz when adequately cooled.
In some perspective, you could say Intel is already doing my idea, since most people don't buy a K or X series CPU and not overclock it. But my point is by blurring the line of how the CPU is supposed to run, that could give Intel a marketing advantage.
I don't think you're understanding my idea. This isn't considered overclocking, and the CPU would be officially running within its specs. 2 years ago, the turbo frequency of a 7700K would be considered an overclock, and yet it is still fully reliable. It isn't hard to push one slightly harder without causing any stability issues. The CPU drops the turbo when it is on the verge of being unstable. The only difference between my idea and what Intel does now is there would be no base frequency (because it is literally meaningless when you account for thermal throttling) and the "turbo speed" would be a little bit higher.
The base clock is what the CPU is rated at for its TDP. It is the baseline specs, which guarantees a certain level of performance, and needs to be specified somewhere. Don't forget that OC is dependent on the CPU (silicon lottery) and the cooling, and Intel cannot sell their products based on what's theoretically possible from their silicon and with third-party coolers and such.
I understand that, but if my idea were to be done then consumers would basically just pick the heatsink and/or PSU that best suits the TDP they're willing to have. Like I said, this would be based on a clock that can be reliably achieved, so basically Intel would take the worst-case scenario of a CPU on liquid cooling, drop the frequency 100Mhz or so, and that would be the maximum speed of the CPU (without overclocking). Silicon lottery is becoming less and less relevant for the average user. I get the impression most 7700Ks can reach 5GHz when adequately cooled.
In some perspective, you could say Intel is already doing my idea, since most people don't buy a K or X series CPU and not overclock it. But my point is by blurring the line of how the CPU is supposed to run, that could give Intel a marketing advantage.
Overclocking is running a CPU beyond spec, and Intel takes no responsibility for it, and for good reason - to do what you are suggesting would mean Intel would be responsible for any damages that happen and would be liable (the lawsuits would be unending!). It would be an absolute disaster for Intel. Also, most people do not overclock, and want reliable and consistent performance for their products - like any product, there needs to be a guarantee in order for consumers to have confidence in a product.
I don't think you're understanding my idea. This isn't considered overclocking, and the CPU would be officially running within its specs. 2 years ago, the turbo frequency of a 7700K would be considered an overclock, and yet it is still fully reliable. It isn't hard to push one slightly harder without causing any stability issues. The CPU drops the turbo when it is on the verge of being unstable. The only difference between my idea and what Intel does now is there would be no base frequency (because it is literally meaningless when you account for thermal throttling) and the "turbo speed" would be a little bit higher.
Embra
Senior Member
Posts: 1292
Joined: 2014-11-19
Senior Member
Posts: 1292
Joined: 2014-11-19
#5455075 Posted on: 07/24/2017 05:56 PM
The vast majority of people buying cpus do not OC at all. There has to be a guaranteed spec min spec.
The vast majority of people buying cpus do not OC at all. There has to be a guaranteed spec min spec.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 6555
Joined: 2012-11-10
But that's the thing - they don't guarantee anything. CPUs will thermal throttle, which is really the inverse of what Turbo Boost does.
The first gen of TB used to be a fantastic and really practical feature, where it would only push 1 or 2 cores past the base frequency. This was nice because the CPU remained at the same wattage while offering more performance for single-threaded tasks. Now, TB applies to all cores so in actuality, it is the base clock and anything below that is technically thermal throttling. At this point it is just a marketing gimmick and an excuse for Intel to ship inadequate heatsinks (AMD used to also do this but their Ryzen heatsinks are legitimately good).