Who's buying Skylake-X? (You may now change your vote)

[Zucker2k]

and how do I know if my OC is stable? I mean, lets say I try a 4.6 OC, but if SKY-X throttles to 4.2 using the common tools to check OC stability, how do I know it will be stable at 4.6?

This is the issue. I am in this "business" since 486 days, and this is the first time this issue happens as far as I can remember

If it passes P95 (AVX version) at 4.2ghz, you'll more than likely be rock stable at 4.6ghz. AVX consumes an abnormally high rate of power hence even at stock clocks, Intel has an in-built AVX-Offset which lowers clocks when running AVX code in order to stay within tdp/power limits.

 

[DrMrLordX]

Hopefully they keep at it and others join in also, because right now it's only when using Prime95 (and with specific settings?) that they can actually reproduce this (as far as I understood it)... It would be interesting if they can reproduce it in actual applications (not sure which would be candidates)...

Linpack and y-cruncher would be my top 2 to test beyond Prime95. And um x265 benchmark maybe?

Enough with all the RYZEN please!

People keep bringing it up, so inevitably it gets responses. FWIW I tried to make it all relateable to Intel platforms (x299 vs Z270 instead of x299 vs AM4, etc).

What exactly can I use USB 3.1 for? Are there 3.1 flash drives? Or is it more for external GPUs?

There's supposed to be 3.1 gen2 flash drives. Not sure if they can sustain data rates that justify the change of interface though. I mean 10 GBps? Okay. There are some . . . other things you're supposed to be able to do over USB 3.1 but it's all kind of hazy. Sometimes I have problems caring about it since I don't need to transfer that much data over USB generally. But apparently someone cares quite a bit.

If it passes P95 (AVX version) at 4.2ghz, you'll more than likely be rock stable at 4.6ghz. AVX consumes an abnormally high rate of power hence even at stock clocks, Intel has an in-built AVX-Offset which lowers clocks when running AVX code in order to stay within tdp/power limits.

That's testable. Wonder if we can get any folks to volunteer?

 

[AdamK47]

If it passes P95 (AVX version) at 4.2ghz, you'll more than likely be rock stable at 4.6ghz. AVX consumes an abnormally high rate of power hence even at stock clocks, Intel has an in-built AVX-Offset which lowers clocks when running AVX code in order to stay within tdp/power limits.

I wouldn't make that assumption. There are many factors that come into play with stability. Saying a Prime95 run at 4.2GHz means 4.6GHz with everything else will be stable isn't correct. A blend of different tests is the best approach. Prime95 is a good start. I'd follow it up with non-AVX test.

 

[Edrick]

Anyone else having issues with Intel Speed Shift (not EIST) with X series CPUs? Turns out my entire m.2 performance issues were caused by Speed Shift down-clocking my bus speed. Only when I run a task in the background am I able to get expected performance on my m.2 drives. Disabling it in the BIOS does not work (ticket open with Gigabyte on that).

 

[LTC8K6]

and how do I know if my OC is stable? I mean, lets say I try a 4.6 OC, but if SKY-X throttles to 4.2 using the common tools to check OC stability, how do I know it will be stable at 4.6?

This is the issue. I am in this "business" since 486 days, and this is the first time this issue happens as far as I can remember

I don't think throttling with "burn-in" type tests is new.

Why not just run the programs you are going to use for a while at the 4.6 clock and see if it works okay?

I would think testing with the programs you are actually going to use makes the most sense.

For analysis and stress testing I use Intel's XTU. If it's stable with that, I call it good.

 

[LTC8K6]

Anyone else having issues with Intel Speed Shift (not EIST) with X series CPUs? Turns out my entire m.2 performance issues were caused by Speed Shift down-clocking my bus speed. Only when I run a task in the background am I able to get expected performance on my m.2 drives. Disabling it in the BIOS does not work (ticket open with Gigabyte on that).

Being Skylake, I would think they would not have Speed Shift, but the Enhanced Speed Step for desktops? Kaby Lake got the newer Speed Shift, and it's really for mobile?

Maybe the KL-X chips have Speed Shift?

EDIT: It's confusing as to which chips have Speed Shift, so the above may be inaccurate.

 

[FIVR]

If it passes P95 (AVX version) at 4.2ghz, you'll more than likely be rock stable at 4.6ghz. AVX consumes an abnormally high rate of power hence even at stock clocks, Intel has an in-built AVX-Offset which lowers clocks when running AVX code in order to stay within tdp/power limits.

Haha... I'd love to see your explanation for how this statement is true. How are you able to decide whether or not a system is stable at 4.6Ghz by running one workload at 4.2Ghz?


I am ever more disappointed in your contributions to this thread. Almost everything you type is either incorrect, inaccurate, or in the case of this post above completely nonsensical.

 

[LTC8K6]

Looks like SL-X is indeed supposed to have Speed Shift. It's not on the ARK pages, but I know those are notoriously unreliable as to specific features of chips.

http://www.anandtech.com/show/11550...-core-i9-7900x-i7-7820x-and-i7-7800x-tested/7

 

[Zucker2k]

I wouldn't make that assumption. There are many factors that come into play with stability. Saying a Prime95 run at 4.2GHz means 4.6GHz with everything else will be stable isn't correct. A blend of different tests is the best approach. Prime95 is a good start. I'd follow it up with non-AVX test.

Point well noted, but I answered the question based on AVX-512 power consumption at 4.2ghz and typical everyday power consumption at 4.6ghz. With the 4600mhz also, certain settings/voltages may need to be tweaked to cater for the 400mhz extra frequency.

 

[Zucker2k]

Haha... I'd love to see your explanation for how this statement is true. How are you able to decide whether or not a system is stable at 4.6Ghz by running one workload at 4.2Ghz?


I am ever more disappointed in your contributions to this thread. Almost everything you type is either incorrect, inaccurate, or in the case of this post above completely nonsensical.

Your cluelessness about AVX loads compared to your typical everyday software is glaring. Take Tahoe's 7820x for example. 4.7ghz stable in everything, only 4.2ghz stable in P95 with AVX. He's running an AVX offset of -500mhz in bios. Clueless much?

Insulting other members is not allowed.
Markfw
Anandtech Moderator

 

[MarkPost]

I don't think throttling with "burn-in" type tests is new.

Why not just run the programs you are going to use for a while at the 4.6 clock and see if it works okay?

I would think testing with the programs you are actually going to use makes the most sense.

For analysis and stress testing I use Intel's XTU. If it's stable with that, I call it good.

well, with what its supossed to be a common OC (not a high one), its the first time I see this thing.

P95 and other tools are intended to be a way to be sure an OC will be stable at "X" speed. I cant (and I dont think anyone can) to use any of my/their common apps (encoding, rendering, image editing or whatever) without being confident system will not crash. Imagine if im doing a large encoding work and it crashes in the middle due to that oc wasnt really stable. Thats why P95 and other tools like P95 are used, to avoid it as much as possible. But it seems with SKY-X thats a no go.

 

[Zucker2k]

well, with what its supossed to be a common OC (not a high one), its the first time I see this thing.

P95 and other tools are intended to be a way to be sure an OC will be stable at "X" speed. I cant (and I dont think anyone can) to use any of my/their common apps (encoding, rendering, image editing or whatever) without being confident system will not crash. Imagine if im doing a large encoding work and it crashes in the middle due to that oc wasnt really stable. Thats why P95 and other tools like P95 are used, to avoid it as much as possible. But it seems with SKY-X thats a no go.

What chip do you plan to get?

Edit: Basically, AVX-512 is the power hungriest code your chip would probably ever encounter. It's an unrealistic, worst-case scenario. Even Intel runs that code with a negative offset in bios. That's why I mentioned that if it's throttling at 4.2ghz running AVX-512 code, you'd more than likely be rock stable at 4.6ghz.

 

[MarkPost]

I have ordered a 7800x and MSI Carbon, hopefully next week will be here

 

[FIVR]

Your cluelessness about AVX loads compared to your typical everyday software is glaring. Take Tahoe's 7820x for example. 4.7ghz stable in everything, only 4.2ghz stable in P95 with AVX. He's running an AVX offset of -500mhz in bios. Clueless much?

So you explanation for how you know any Skylake-X processor is capable of running at 4.6Ghz if it runs P95 at 4.2Ghz is somebody told you they own one specific 7820x that behaves that way?

LOL. Keep calling people "clueless", it suits you.

 

[Zucker2k]

I have ordered a 7800x and MSI Carbon, hopefully next week will be here

With a 7800x, 4.6ghz should be a breeze. You should even be stable with the non-avx version of P95 at those clocks. The AVX-512 version of P95, however, I'm not so sure. So, unless you plan to run anything with AVX-512 code besides stress-testing, I wouldn't worry about being stable at 4.6ghz for the 7800x. That chip is clocking as high as 4.8ghz stable.

 

[Zucker2k]

So you explanation for how you know any Skylake-X processor is capable of running at 4.6Ghz if it runs P95 at 4.2Ghz is somebody told you they own one specific 7820x that behaves that way?

LOL. Keep calling people "clueless", it suits you.

Okay, so you have a comprehension problem. Take your issue somewhere else.

Insulting other members is not allowed.
Markfw
Anandtech Moderator

 

[FIVR]

Okay, so you have a comprehension problem. Take your issue somewhere else.

Does your brilliant discovery also follow for Ryzen processors? If a Ryzen can rune P95 stable at 3.9Ghze, then is it able to run any other workload at 4.3Ghz?

I guess it also means that any Skylake - X that runs P95 stable at 4.4Ghz is capable of 4.8Ghz stable? Or does the magic frequency increase only apply to processors running P95 at 4.2-4.3Ghz?


I'm just trying to gain some of that great knowledge you have about processors, LOL!

 

[AdamK47]

I think some people have misconstrued the real intended purpose of the AVX offset. It's to help keep the processor within thermal and power envelopes. Stability at a reduced offset is only a byproduct of this. That same CPU may be stable without any offset at a given voltage and speed. Using the offset should not be read into as a guarantee for stability.

For me, I would stability test overclocks with AVX enabled at a safe speed and voltage level without any AVX offset. With those settings I would run AVX tests and non AVX tests to ensure stability. Only after determining stability would I then set an AVX offset from that given speed to ensure power and temperatures fall to realistic everyday usage levels. That's just me though. I don't get greedy with overclocks. ~4.5GHz with as low a voltage as necessary would suffice.

 

[lolfail9001]

Does your brilliant discovery also follow for Ryzen processors? If a Ryzen can rune P95 stable at 3.9Ghze, then is it able to run any other workload at 4.3Ghz?

Well, you seriously claimed that 1800X would OC to 4.3Ghz, so i guess it does.
Wait, it does not?

I'm just trying to gain some of that great knowledge you have about processors, LOL!

I guess you really need that knowledge, then.

you'd more than likely be rock stable at 4.6ghz.

Does not work that way, sadly, AVX stability and non-AVX stability are apples and oranges. So, just fire up both AVX and non-AVX versions of prime for the stability testing purposes and use the damn AVX offset (it's not like you will be any slower than Ryzen in AVX workloads even with 2Ghz offset lul).

 

[FIVR]

Well, you seriously claimed that 1800X would OC to 4.3Ghz, so i guess it does.
Wait, it does not?

Not much of an issue that it fell short, now that Skylake-X can barely even run at stock clocks without drawing 200+ Watts.

I guess you really need that knowledge, then.


Does not work that way, sadly, AVX stability and non-AVX stability are apples and oranges. So, just fire up both AVX and non-AVX versions of prime for the stability testing purposes and use the damn AVX offset (it's not like you will be any slower than Ryzen in AVX workloads even with 2Ghz offset lul).

How long do you think you'll last before you get banned again, failtroll9001?

 

[LTC8K6]

Not much of an issue that it fell short, now that Skylake-X can barely even run at stock clocks without drawing 200+ Watts.


How long do you think you'll last before you get banned again, failtroll9001?

Stop swallowing the hook...

 

[TahoeDust]

Everytime I come back to this thread, it get muddier and muddier.

I changed ram kits to some Samsung B-Die Gskill...

 

[lolfail9001]

Not much of an issue that it fell short, now that Skylake-X can barely even run at stock clocks without drawing 200+ Watts.

You've mistyped 140. Happens with trolls i know.

How long do you think you'll last before you get banned again, failtroll9001?

How long will you?

Insulting other members is not allowed.
Markfw
Anandtech Moderator

 

[tamz_msc]

If those 4000MHz+ kits really work in quad-channel mode on X299, then we might have aida64 memory bandwidth overtaking L3 bandwidth on these chips without overclocks.

 

[TahoeDust]

With the new kit, my little 8 core broke 2100...