Skylake-X (Core i9) - Lineup, Specifications and Reviews!

Nice editorializing.

The Article said:
That doesn't really explain why the 7800X was just flat out slow by comparison for quite a few of the games tested. The likely reason for this is down to Intel restructuring the cache hierarchy. Compared to the 7700K, the 7800X has quadrupled the L2 cache per core while the shared L3 has been reduced by just over 30% per core. It's believed these changes combined with the way this new cache works makes Skylake-X more suited for server-related tasks and less efficient when it comes to things such as gaming, and that's certainly what we're seeing here.

...

If you're a gamer, you should get the 7700K or look to AMD's Ryzen lineup.
Nowhere in your link do they state that Skylake-X is "closer to Ryzen in gaming than Kabylake (X)." Never mind the fact that they compared it to a Kabylake chip, the 7700K, not a Kabylake-X chip.
 
Nice editorializing.


Nowhere in your link do they state that Skylake-X is "closer to Ryzen in gaming than Kabylake (X)." Never mind the fact that they compared it to a Kabylake chip, the 7700K, not a Kabylake-X chip.

Its a regression in performance, the 7700K should not be faster then the 7800x and yet it is.
 
Its a regression in performance, the 7700K should not be faster then the 7800x and yet it is.
Did you read the part I quoted? And again, if you're buying a HEDT processor just for gaming then you need to reexamine your priorities. I mean, unless you're really starving for PCI-E lanes by running a lot of PCI-E SSD or something.
 
Did you read the part I quoted? And again, if you're buying a HEDT processor just for gaming then you need to reexamine your priorities. I mean, unless you're really starving for PCI-E lanes by running a lot of PCI-E SSD or something.

A 6 core processor is not a HEDT setup man. Just cause Intel likes to rip people off does not make it a HEDT. Your 6 core is for the guy that likes to game more often then using it for productivity except the 7800x is a large regression on gaming performance despite the speed being almost the same. They clocked it up and it did not matter the 7800x is a disappointment there. Just sounds like excuses to me why it has regressed Armenius, hell Ryzen may not need a upgrade at the rate Intel is going.

Average.png
 
Why not? Uncore is clocked 900Mhz higher, base clock is 700Mhz higher and turbo clock is 500Mhz higher.

If you check hardwareunboxed, even at 4.7ghz it was slower than the stock 7700K. But I saw a review of a 7900X that was 20%+ faster than one of the engineering samples that was given to a reviewer. But the retail sample was delidded, so it’s also possible the stock performance is limited by heat transfer issues. Either way, no point going X299 for strictly gaming.
 
Will be interesting to see what comes of the larger core count chips such as the 16 and 18 core models. Especially since the mobos on the market now weren't initially designed for them, we might see a v2 of some of them.
 

Not even close. This article is "Intel vs Intel". There is no RyZen reviewed. The only reason why they barely mention RyZen in the conclusions is for price, not performance.

RyZen numbers are found in the Skylake-X review

https://www.techspot.com/review/1433-intel-core-i9-core-i7-skylake-x/page3.html

where you can find top 8-core Ryzen loosing to six-core Skylake even in games more favorable to AMD

Ashes.png


http://www.amd.com/en-us/markets/game/featured/ashes-of-the-singularity#
https://www.extremetech.com/gaming/...update-substantially-boosts-ryzen-performance
 
Skylake-X is not designed for gaming and still it is only 8% behind the king of gaming CPUs. Note as well that OC the mesh in SKL-X gives huge double digit performance improvements in games.

Hmmm that sounds interesting. Maybe this is a dumb question, but how do you OC the mesh?
 
In the bios. Increase frequency and add voltage as needed. Just like overclocking core. Factory is 2400MHz. In my case 3200MHz was easily attainable.

Would you mind running some benchmarks?
 
Not even close. This article is "Intel vs Intel". There is no RyZen reviewed. The only reason why they barely mention RyZen in the conclusions is for price, not performance.

RyZen numbers are found in the Skylake-X review

https://www.techspot.com/review/1433-intel-core-i9-core-i7-skylake-x/page3.html

where you can find top 8-core Ryzen loosing to six-core Skylake even in games more favorable to AMD


http://www.amd.com/en-us/markets/game/featured/ashes-of-the-singularity#
https://www.extremetech.com/gaming/...update-substantially-boosts-ryzen-performance


Really AoS? Great example. Lets look at Techspot quote fro a real game:

"I triple checked the numbers though and they are indeed accurate. Interestingly this is reminiscent of what we saw when comparing the, the 8-core Ryzen 7 1800X for example was much much slower than the 7700K. In fact, given what I'm seeing here, the 1800X might actually be faster than the overclocked 7800X in this title so that's something I'm keen to check out soon."

I don't remember x99 losing to mainstream Intel processors this much when comparing similiar frequencies. No reason the HEDT should have to make so many comprimises.
 
Would you mind running some benchmarks?
Which ones would you like to see?

I've got my 7820x running at 4.6/4.7/4.8 (8c/4c/2c) with mesh OC to 3.0Ghz.

Here are a few I just ran with my 1080 Ti at stock (100% power, +0/+0):
3DMark Time Spy 10014 (graphics 9843, CPU 11109)
3DMark Firestrike Normal 22631 (graphics 27724, physics 24312)
3DMark Firestrike Ultra 7133 (graphics 7018, physics 23504)
Unigine Superposition 1080p Medium 18210
Unigine Superposition 1080p Extreme 5812
Unigine Superposition 4K Optimized 9247

EDIT: I set the mesh back to stock, 2.4Ghz, and re-ran the benchmarks to help isolate it:
3DMark Time Spy 9960 (graphics 9854, CPU 10611)
3DMark Firestrike Normal 22089 (graphics 27341, physics 24032)
3DMark Firestrike Ultra 7113 (graphics 6985, physics 23984)
Unigine Superposition 1080p Medium 18033
Unigine Superposition 1080p Extreme 5792
Unigine Superposition 4K Optimized 9239

Now back to BF1 multiplayer, where this thing is a beast.
 
Last edited:
10% to 20% is fairly impressive. What is the general speed of the cpu and the ram?
 
Which ones would you like to see?

I've got my 7820x running at 4.6/4.7/4.8 (8c/4c/2c) with mesh OC to 3.0Ghz.

Here are a few I just ran with my 1080 Ti at stock (100% power, +0/+0):
3DMark Time Spy 10014 (graphics 9843, CPU 11109)
3DMark Firestrike Normal 22631 (graphics 27724, physics 24312)
3DMark Firestrike Ultra 7133 (graphics 7018, physics 23504)
Unigine Superposition 1080p Medium 18210
Unigine Superposition 1080p Extreme 5812
Unigine Superposition 4K Optimized 9247

EDIT: I set the mesh back to stock, 2.4Ghz, and re-ran the benchmarks to help isolate it:
3DMark Time Spy 9960 (graphics 9854, CPU 10611)
3DMark Firestrike Normal 22089 (graphics 27341, physics 24032)
3DMark Firestrike Ultra 7113 (graphics 6985, physics 23984)
Unigine Superposition 1080p Medium 18033
Unigine Superposition 1080p Extreme 5792
Unigine Superposition 4K Optimized 9239

Now back to BF1 multiplayer, where this thing is a beast.

Not bad, looks like about the same bump Ryzen gets with higher clock speeds with Ram. Would be nice if I could control the Fabric speed like the Mesh. It tends to have the greatest effect at low resolutions, once your at 1440p and above it tends to have little effect on fps or in this case benchmark scores. Nice to see the effect it has tho.
 
Thats pretty slow base clock, it's going to be a problem for them with their current process as the core count goes up it favors AMD

I am not so sure that the 2.9GHz is really an issue. I mean with the higher IPC of the i9 would that slower or faster per thread than a 3.4GHz TR? I would expect it to be still a little faster per core than the 16C TR. But the 16C TR will clearly be better if you can use 32 threads.
 
I am not so sure that the 2.9GHz is really an issue. I mean with the higher IPC of the i9 would that slower or faster per thread than a 3.4GHz TR? I would expect it to be still a little faster per core than the 16C TR. But the 16C TR will clearly be better if you can use 32 threads.

TR appears to be ~10 to 15% slower per clock. So they will likely be close to even.
 
TR appears to be ~10 to 15% slower per clock. So they will likely be close to even.

Except the problem is the 7800x is showing a drop in performance of around 10% in some benchmarks, not all. So your talking about a 5% gap or less between AMD and Intel. The 7900X has issues in some benchmarks as well. Intel is using server chips and using them in a normal desktop way and it's hurting them a bit. Will be interesting to see how the benchmarks end up tho.
 
At such a low base clock, I don't see the 12 core being any faster multi threaded than the 10 core while being a good bit slower at single threaded use. So where is the gain?
 
Potentially unpopular opinion, but I kinda feel like AMD and Intel are getting ahead of themselves with these 12+ core CPUs. HEDT is already niche, but finding a buyer for a 16 core who isn't already better served by the server platforms seems like it's super duper niche. On top of that, it seems like with all the heat struggles we're already seeing with the 10 core, I feel like the 16 cores are going to have to have too low of clock speeds to be interesting to almost anybody (and again, CPUs with tons of cores and low clocks already exist in the server land, so why recreate them?) Maybe I'm missing something.
 
Potentially unpopular opinion, but I kinda feel like AMD and Intel are getting ahead of themselves with these 12+ core CPUs. HEDT is already niche, but finding a buyer for a 16 core who isn't already better served by the server platforms seems like it's super duper niche. On top of that, it seems like with all the heat struggles we're already seeing with the 10 core, I feel like the 16 cores are going to have to have too low of clock speeds to be interesting to almost anybody (and again, CPUs with tons of cores and low clocks already exist in the server land, so why recreate them?) Maybe I'm missing something.

ThreadRipper includes 64 PCIE lanes, that's a big deal.
 
ThreadRipper includes 64 PCIE lanes, that's a big deal.

For those few who need it. Most users will have no need for that.

I could use the cores in my medical imaging research work but have absolutely no need for that many lanes on a workstation. A server possibly but that's a lot of raid cards / HBAs.
 
Last edited:
For those few who need it. Most users will have no need for that.

I could use the cores in my medical imaging research work but have absolutely no need for that many lanes on a workstation. A server possibly.
Sure, but others may. Capture cards / Audio cards / More Nics / .... other things (admittedly im not the target market here either).

I dont think ThreadRipper or SKL-X are gonna set the world on fire, but I do think a sub $3000 workstation with such capabilities have a place.
 
I think 'yall are missing a bigger picture here comparing the 12 core i9 with the 12 core RTR.

Even if the RTR is - say - 10 % slower in ALL tasks, its still $799, versus the (likely) $1500+ i9.
 
I think 'yall are missing a bigger picture here comparing the 12 core i9 with the 12 core RTR.

Even if the RTR is - say - 10 % slower in ALL tasks, its still $799, versus the (likely) $1500+ i9.

You missing cost of mobos, memory,...

Those chips don't target games, but professional people. For some of us getting the work done on 20% less time compensates for the acquisition price gap.
 
People are indeed missing the picture, and its called turbo clocks ;)

Ok even with Turbo Core, low thread workloads MIGHT be as fast as the 7900x while still being about the same on multi thread with the lower freq. So again, what is the point?
 
Ok even with Turbo Core, low thread workloads MIGHT be as fast as the 7900x while still being about the same on multi thread with the lower freq. So again, what is the point?

Could you elaborate? Else its like claiming a 7820X will be as fast as a 7900X.

What's the all core turbo on the 7920X? :)
 
Last edited:
Doesn’t look like the Mesh OC did much on a 7800X. Maybe it helps with the 7900X or you need to OC it to 3.2ghz, but so far it sounds like this isn’t the magic fix.

Can’t link it right now but hardwareunboxed just did a video on it. Basically it helped in some games, but once you overclocked the CPU it was within the margin of error for the most part. 7700K still steamrolled it in most games.
 
Well, we know that 7920X is going to cost 1189$ and its base frequency is going to be 2.9Ghz... we can only estimate turbo frequencies but I'm pretty sure the all core turbo frequency is going to be lower on the 7920X compared to 7900X. Turbo 3.0 (2 cores) freq might actually be the same.
 
That's for damn sure! The 7900x is already running into thermal and power limits. No way are you going to get a higher all core by adding more cores on this platform.

Can't wait until the 2.0 Ghz 18 core!
 
Back
Top