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

cageymaru said:
London 1666
http://www.tomshardware.de/intel-ku...ottling-ubertakten,testberichte-242373-4.html

Rome 64 AD
aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9FL0QvNjkxMjg1L29yaWdpbmFsLzAzLTI1MC1XYXR0cy5wbmc=


Chicago 1871

Click to expand...


English version

http://www.tomshardware.com/reviews/-intel-skylake-x-overclocking-thermal-issues,5117.html
 
bwang said:
The problem with AVX/FMA/AVX512 is that while they help generate amazing synthetic scores, they aren't very useful in practice. It's not even that applications haven't been recoded to take advantage of the new extensions, but rather FP throughput is, in almost all real-world cases, limited by bandwidth and/or latency and not raw FPU throughput. Linpack benefits from it (large amounts of data reuse, easy caching), but nothing uses a single giant LU decomposition, limiting its use to mostly nation-scale benchmarking. Signal processing problems using Fourier transforms similar to Prime95's Small FFT benchmark could benefit from it, but I have yet to see signal processing code that benefits from AVX (CERN's benchmarking a few years ago with their own collider analysis codes showed that there was little speedup going beyond the 128-bit vector units that SSE provides).
Conventional engineering problems certainly do not benefit from newer than SSE, if that. FEA typically uses iterative sparse matrix solvers, and the fundamental structure of a sparse matrix (you don't know which elements are nonzero or where they are) makes it hard to cache. Renderers don't seem to take advantage of AVX512 much either (we don't see the 7900X being 2x faster than the 7820K despite having 25% more cores and twice the FMA's).

Now this would be all fine and dandy if AVX were free. Quite frankly, on a stock-clocked processor it is - the extra power the FP units need is offset by lower AVX clocks, and in the end in the same thermal envelope you get almost twice the FP throughput in the best case, and minimal loss in the worst case (with an overly aggressive implementation of AVX offset, some applications which contain AVX code but does not benefit from it will see slightly decreased performance).
Unfortunately, from an enthusiast point of view AVX is a pain in the ass when it comes to overclocking. Nothing remotely consumer benefits from it (and this includes "workstation" apps such as Photoshop/Blender/Keyshot/Handbrake, etc), but you can't very well have a computer which goes and crashes under heavy AVX load, even if you don't expect such a load. AVX offset solves the problem, but makes stability testing hard - most of the existing power viruses used to test for stability and thermal headroom trigger the offset, making it difficult to find a representative heavy workload for accelerated stability testing.
Click to expand...

The memory wall is basic computer architecture problem, not a problem exclusive to the SIMD approach. And for latency-bound workloads, using AVX or other SIMD approach on a CPU is much better than using GPGPU due to huge interconnect latencies.

There are good number now of AVX512-based supercomputers, and they run AVX512 code to crunch numbers faster than AVX2, SSE or GPGPUs.

AVX512 providing ~2x speedup over AVX2

https://indico.cern.ch/event/432527...tachments/1320075/1979329/GeantV_ICHEP_16.pdf

Other advantages of AVX512 over AVX2 and SSE

https://arxiv.org/pdf/1704.08579.pdf

https://www.hpcwire.com/2017/06/29/reinders-avx-512-may-hidden-gem-intel-xeon-scalable-processors/

And so on. I have a friend that has developed a 3D rendering engine that uses AVX instructions. And Google has been using AVX512 on servers for a while.
 
Thermals can be misleading, for starters most reviewers us a test bed instead of a chassis, if you test in a chassis, not all have the same air flow and heat dissipation for example my phanteks is piss poor at moving air and thermal build up is real, in summer where my ambient can hit 35 degrees Celsius I have to open the chassis or inside two minutes of battlefield and the system will black screen lock up, seen my 4790K hit 90+ with a 120m Corsair AIO not overclocked. Which is the last point, they test in air conditioned labs with test benches with lower ambient and air dissipation, this is not a realistic condition.

Do I think these CPU's will run hot, yes very likely they will. If you life in Denmark where it can barely pass our winter temps in summer you will probably be able to cool it better than a person that lives slap bang in the middle of Arizona, or Dubai or in most southern hemisphere countries.
 
OrangeKhrush said:
Thermals can be misleading, for starters most reviewers us a test bed instead of a chassis, if you test in a chassis, not all have the same air flow and heat dissipation for example my phanteks is piss poor at moving air and thermal build up is real, in summer where my ambient can hit 35 degrees Celsius I have to open the chassis or inside two minutes of battlefield and the system will black screen lock up, seen my 4790K hit 90+ with a 120m Corsair AIO not overclocked. Which is the last point, they test in air conditioned labs with test benches with lower ambient and air dissipation, this is not a realistic condition.

Do I think these CPU's will run hot, yes very likely they will. If you life in Denmark where it can barely pass our winter temps in summer you will probably be able to cool it better than a person that lives slap bang in the middle of Arizona, or Dubai or in most southern hemisphere countries.
Click to expand...

I remember Jayz2cents compared his open bench to enclosed case and found there was no difference between the two in terms of thermal performance, this was also done with dual GPUs to really push it.
Cheers
 
OrangeKhrush said:
Do I think these CPU's will run hot, yes very likely they will. If you life in Denmark where it can barely pass our winter temps in summer you will probably be able to cool it better than a person that lives slap bang in the middle of Arizona, or Dubai or in most southern hemisphere countries.
Click to expand...

I am sure its easier to cool in those 2 locations than Denmark in the summer due to the usage of AC ;)

Now had you used Portugal, Spain, Greece or Italy then it would be likely.
 
CSI_PC said:
I remember Jayz2cents compared his open bench to enclosed case and found there was no difference between the two in terms of thermal performance, this was also done with dual GPUs to really push it.
Cheers
Click to expand...

then jaytwocents posted a video on how he is not using his custom cooled ryzen rig because thermal performance was terrible in his phanteks and he is to lazy to take it apart
 
Hmm I would really like to see the [H] review and perhaps a cooler roundup to see which coolers can cut the mustard with overclocking. That would be really usefull :)
 
Intel officially state that poor gaming performance as shown by hardware unboxed is down to Intel "mesh" affecting performance. Much like Ryzen Infinity Fabric it looks like Intel are also suffering high latency.

Jay Jardin in WCCF troll section posted AIDA scores, the L3 latency was twice that of Ryzen at over 20ms.

Of course all the intel fans are "we will wait for patches". Completely unimpressive release.

The same JayJardin posted a Cinebench MT score of 2126pts at 4.8Ghz with 4000mhz RAM speeds on the 7820X. A 1800X at 4.8Ghz would score 2130 in principal to so it is not impressive at all given the cost disparity. Where I live after import tax and reseller mark up the boards are 7-12K while the 7820X costs about 22K relative to Ryzen 1800X which comes in at 6K and a top end board at 4K its not good value. My 4790K is still looking so good.
 
get


d9767028a962728bbf5107e4bdb95b6b50a5213042e7a9286265edf762711d5b.png


It is sad that a company like Intel needs to cherry pick its benchmarks and play in an area that AMD is only re entering as a saving grace for inability to advance. I like how a very expensive Xeon is compared to a 400 dollar mainstream chip. Kudos

Anyways Anandtech has their Epyc vs Xeon review and there is legitimate competition whichever way you want to slice it. Amen to competition.
 
OrangeKhrush said:
get


d9767028a962728bbf5107e4bdb95b6b50a5213042e7a9286265edf762711d5b.png


It is sad that a company like Intel needs to cherry pick its benchmarks and play in an area that AMD is only re entering as a saving grace for inability to advance. I like how a very expensive Xeon is compared to a 400 dollar mainstream chip. Kudos

Anyways Anandtech has their Epyc vs Xeon review and there is legitimate competition whichever way you want to slice it. Amen to competition.
Click to expand...

You mean the review that doesn't match other reviews at all and even contradicts their own SKL-X review? Ye, lets put a lot of trust into that one. ;)

http://www.tomshardware.com/reviews/intel-xeon-platinum-8176-scalable-cpu,5120.html
https://www.servethehome.com/quad-intel-xeon-platinum-8180-initial-benchmarks/
 
OrangeKhrush said:
it is an intel slide not review

434cb2869c9c7836f41ae24f34584f5124d61ff4b3e630b6ff5d4b16913ac366.jpg


AMD glued dies together apparently.
Click to expand...

We all know that don't we? Don't tell me its a sudden surprise for you.

And you mentioned the AMD/Anandtech review. ;)

Lots of things falling apart, no better SMT on Zen and what else that have tried to push the PR machine.

Talking about SMT, when will AMD fix their broken SMT? And did they fix the broken FMA units yet? Or how about the VME bug?
 
Shintai said:
We all know that don't we? Don't tell me its a sudden surprise for you.

And you mentioned the AMD/Anandtech review. ;)

Lots of things falling apart, no better SMT on Zen and what else that have tried to push the PR machine.

Talking about SMT, when will AMD fix their broken SMT?
Click to expand...

You seem to be a know it all, you tell me? actually make it public domain on twitch and youtube that AMD glues stuff together, then come explain the L3 latency at 20+, then keep up the damage control.

I know you have undying love for Intel like they actually care for your needs, I know you have your little therepy sessions in these forums where the same three get together and allay their inferiority complexes about chips you can't afford anyway and talk about how it makes your feel better. I am okay with the group support therepy to feel good.
 
OrangeKhrush said:
You seem to be a know it all, you tell me? actually make it public domain on twitch and youtube that AMD glues stuff together, then come explain the L3 latency at 20+, then keep up the damage control.

I know you have undying love for Intel like they actually care for your needs, I know you have your little therepy sessions in these forums where the same three get together and allay their inferiority complexes about chips you can't afford anyway and talk about how it makes your feel better. I am okay with the group support therepy to feel good.
Click to expand...

I see you are out of arguments already ;)
 
OrangeKhrush said:
Intel officially state that poor gaming performance as shown by hardware unboxed is down to Intel "mesh" affecting performance. Much like Ryzen Infinity Fabric it looks like Intel are also suffering high latency.

Jay Jardin in WCCF troll section posted AIDA scores, the L3 latency was twice that of Ryzen at over 20ms.

Of course all the intel fans are "we will wait for patches". Completely unimpressive release.

The same JayJardin posted a Cinebench MT score of 2126pts at 4.8Ghz with 4000mhz RAM speeds on the 7820X. A 1800X at 4.8Ghz would score 2130 in principal to so it is not impressive at all given the cost disparity. Where I live after import tax and reseller mark up the boards are 7-12K while the 7820X costs about 22K relative to Ryzen 1800X which comes in at 6K and a top end board at 4K its not good value. My 4790K is still looking so good.
Click to expand...
Show me an 1800X that can hit 4.8 GHz.
 
Show me a Skylake X with solder that can hit those speeds with good thermals unless you plan on delidding a $500+ CPU.
 
Rvenger said:
Show me a Skylake X with solder that can hit those speeds with good thermals unless you plan on delidding a $500+ CPU.
Click to expand...

I have mine at 4.7ghz and maxes out at 80c prime95 fft yet most people don't have an overkill custom water loop like mine lol
 
OrangeKhrush said:
Intel officially state that poor gaming performance as shown by hardware unboxed is down to Intel "mesh" affecting performance. Much like Ryzen Infinity Fabric it looks like Intel are also suffering high latency.

Jay Jardin in WCCF troll section posted AIDA scores, the L3 latency was twice that of Ryzen at over 20ms.

Of course all the intel fans are "we will wait for patches". Completely unimpressive release.
Click to expand...

Initial Skylake-X reviews used buggy early BIOSes that affected latency, but that was solved weeks ago:

The issue is that the normal 7900K results were done on older BIOS versions, while the new ones are done on the latest BIOS versions that support Turbo 3 without any software requirements. You will see results in line with the better performing 7900X results, but I do know some other media who were getting the same low gaming scores I was, and that was because Turbo 3 wasn't working. Memory increase from 2133Mhz to 2666Mhz with the same timings also makes a difference in the gaming results. Ashes of Singularity is one of our outliers, but that is most likely because they have to optimize the code for the CPU, just like had to be done for Ryzen.
Click to expand...


8225_41_intel-core-i9-7900x-series-skylake-cpu-review.png

8225_38_intel-core-i9-7900x-series-skylake-cpu-review.png



Latest reviews already used final BIOS:

Hitman.png

Ashes.png


Note the relative bad performance of the R7-1800X, which is losing to six-core Skylake. RyZen users continue awaiting to those promised "game patches" and "BIOS fixes"...


OrangeKhrush said:
The same JayJardin posted a Cinebench MT score of 2126pts at 4.8Ghz with 4000mhz RAM speeds on the 7820X. A 1800X at 4.8Ghz would score 2130 in principal to so it is not impressive at all given the cost disparity. Where I live after import tax and reseller mark up the boards are 7-12K while the 7820X costs about 22K relative to Ryzen 1800X which comes in at 6K and a top end board at 4K its not good value. My 4790K is still looking so good.
Click to expand...

8-core RyZen is slower than 8-core Skylake in CB15MT, both at stock settings

87127.png


But whereas Intel chips can be pushed to 4.8GHz as Jay said to you, no known RyZen chip can get close to 4.5GHz with ordinary cooling, making your RyZen-at-4.8GHz-score imaginary.

Let us take a look at Blender, where six-core Skylake beats 8-core Zen

87126.png


or let us take a look at WinRAR

87128.png
 
Last edited:
OrangeKhrush said:
get


It is sad that a company like Intel needs to cherry pick its benchmarks and play in an area that AMD is only re entering as a saving grace for inability to advance. I like how a very expensive Xeon is compared to a 400 dollar mainstream chip. Kudos.
Click to expand...

They aren't comparing chips. Intel is comparing the implementations of SMT on each microarchitecture. The implementation of SMT used in RyZEn chips is the same implementation used in EPYC chips. Note that EPYC uses the same Zeppelin dies than RyZen.

OrangeKhrush said:
Anyways Anandtech has their Epyc vs Xeon review and there is legitimate competition whichever way you want to slice it. Amen to competition.
Click to expand...

That biased AT review that disagrees with every review of the same chip, and bring us a picture of the Skylake muarch that even disagrees with AT review of Skylake-X? LOL
 
Last edited:
OrangeKhrush said:
434cb2869c9c7836f41ae24f34584f5124d61ff4b3e630b6ff5d4b16913ac366.jpg


AMD glued dies together apparently.
Click to expand...

That is all well-known now. In fact, some of us predicted that diagram on the righ hand side of the slide many years ago. Here you have a similar four-die diagram that I posted then for AMD

I also explained before that the die-die latency on Naples was going to be about one-order of magnitude higher than the CCX-CCX latency. But this is not all the history. Thanks to infinity hype, those dual-socket EPYC platforms have worse socket-socket latencies:

Putting this into perspective, Intel actually has inter-socket idle latency that is better than our AMD EPYC 7601 system with DDR4-2400 is currently putting out. That is a phenomenal result and will help explain some of the performance findings we have later.
Click to expand...
 
Ok, if Kyle slams intel in his review, you guys will be calling him biased. Maybe someone needs to look in the damn mirror.
 
OrangeKhrush said:
Intel officially state that poor gaming performance as shown by hardware unboxed is down to Intel "mesh" affecting performance. Much like Ryzen Infinity Fabric it looks like Intel are also suffering high latency.

Jay Jardin in WCCF troll section posted AIDA scores, the L3 latency was twice that of Ryzen at over 20ms.

Of course all the intel fans are "we will wait for patches". Completely unimpressive release.
Click to expand...
Cwqpeo7.jpg


Single thread performance is equal or better than 7700k clock for clock...
6dBLDmR.jpg
 
Rvenger said:
Ok, if Kyle slams intel in his review, you guys will be calling him biased. Maybe someone needs to look in the damn mirror.
Click to expand...

Oh cammon! Don't compare a nice neutral site like this with a complete biased site as AT, where the mods are derailing technical threads about Intel products with posts about AMD shares and finances, and then warning both in public and in private messages to anyone disagreeing with them.
 
Skyblue said:
Hmm I would really like to see the [H] review and perhaps a cooler roundup to see which coolers can cut the mustard with overclocking. That would be really usefull :)
Click to expand...
Kyle said that you should keep your expectations low. So on the positive side, he did not say Skylake X sucks.
 
bezant said:
what about HWMonitor
Click to expand...
Yeah. HWMonitor gets it right. It is what I have used througout most of the overclocking process. I just screenshot with CPU-z as it seems to be the standard for clock/memory settings verifications.
 
juanrga said:
Initial Skylake-X reviews used buggy early BIOSes that affected latency, but that was solved weeks ago:




8225_41_intel-core-i9-7900x-series-skylake-cpu-review.png

8225_38_intel-core-i9-7900x-series-skylake-cpu-review.png



Latest reviews already used final BIOS:

Hitman.png

Ashes.png


Note the relative bad performance of the R7-1800X, which is losing to six-core Skylake. RyZen users continue awaiting to those promised "game patches" and "BIOS fixes"...




8-core RyZen is slower than 8-core Skylake in CB15MT, both at stock settings

87127.png


But whereas Intel chips can be pushed to 4.8GHz as Jay said to you, no known RyZen chip can get close to 4.5GHz with ordinary cooling, making your RyZen-at-4.8GHz-score imaginary.

Let us take a look at Blender, where six-core Skylake beats 8-core Zen

87126.png


or let us take a look at WinRAR

87128.png
Click to expand...


So all in all a company that has node advantage, fab advantage and still not all that fabulous, you have to pay so much more

Imaginary possibly but since you compare scores at similar clocks and Cinebench being very linear and scaleable it is not hard to calculate in principal. Anandtech database often contridicts their review scores and that is largely down to cheating numbers or turning off throttling in reviews pushing overclocks and posting numbers as stock. the database numbers are pure base clock numbers run by guys that varify scores on HWBOT.

Seen Jay post a score at 4.8Ghz scoring 2126 so i will accept that as valid given there is window of play inside a 50pt swing depending on the setup and tweakings. Similarly i have seen a Fosmark post a Ryzen 1800X score just shy of 1900marks at 4.2Ghz with 3000mhz on the RAM. All things considered vs BDW E and Ryzen SL X is not that ground breaking at all.
 
n=1 said:
https://www.techspot.com/review/1445-core-i7-7800x-vs-7700k/

Average.png


AverageSlide.png


(ignore the "13% faster @ 1080p, that's simply the average. Chart actually shows percentage 7700K is faster than 7800X by game)

If gaming is all you do, you're still (much) better off buying Z270 over X299.
Click to expand...

I agree that the 7700k is a better choice for ONLY gaming. The 7800X is the preferred option for gaming AND doing pro stuff.

Said that, I don't know from where they got the 13%. In the first image, the average performance gap is 10% on stock settings and 8% with overclocks.
 

amd just announced threadripper pricing

$799 for 12 core
$999 for 16 core..
intel price drop inc?
 
I see all the usual suspects are well at home here, circlejerking like champs. Carry on boys.
 
You must log in or register to reply here.
Back
Top