Intel Skylake-X Review Core i9 7900X

Single thread performance on part with kabylake (at lower clocks)
313f0f1a-3492-4368-a3f3-16915606fa46.png


32351b36-029e-4c8e-8e32-9584e227d1d6.png
 
They said the BIOS is still immature, hence the lower 3D performance.

Yeah was just going to point out they literally saw a 150% increase in performance in VRMark by updating bios (from 4k to 10k score)
 
Looks like voltage needs to stay under 1.3 volts or thermals get out of control and it's sucking some juice for a Intel cpu. 4.7 ghz on 10 cores is pretty good tho but I dont think the 12 core and higher will be hitting that overclock based on the thermals and wattage this chip is using.
 
Looks like voltage needs to stay under 1.3 volts or thermals get out of control and it's sucking some juice for a Intel cpu. 4.7 ghz on 10 cores is pretty good tho but I dont think the 12 core and higher will be hitting that overclock based on the thermals and wattage this chip is using.
4.7/4.8 GHz is very doable, sub 70c with AIO at something like 1.22v based on the couple reviews I've seen and some tech retail sources I've spoken to
 
Just heads-up,
worth remembering if they are releasing a review now they must had borrowed some kind of setup from someone, which means we do not know the state of the CPU and motherboard, including BIOS versions.
This influence may be seen when comparing Bit-Tech who also have a loaner from someone to Hexus.

Cheers
 
Just heads-up,
worth remembering if they are releasing a review now they must had borrowed some kind of setup from someone, which means we do not know the state of the CPU and motherboard, including BIOS versions.
This influence may be seen when comparing Bit-Tech who also have a loaner from someone to Hexus.

Cheers
same applies to any TR leaks we see as well. Still, SKYL looking good
 
Hello threadripper, meet Skylake -X


TR: https://browser.primatelabs.com/v4/cpu/3108993

SKYX:

https://browser.primatelabs.com/v4/cpu/3117213

usual grain of salt for geekbench results

Are you comparing single-thread performance there? Or was those just for kicks?

Let me help you: random 1700x result: https://browser.primatelabs.com/v4/cpu/3135075

I guess this means TR is going to be only as fast as 1700x even though it has double the cores / threads. But we all know that only AMD can achieve this kind of feat.

Edit: And no, I'm not attacking here... just curious and bit drunk :D
 
same applies to any TR leaks we see as well. Still, SKYL looking good
Hehe I put TR much lower down with the amount of information out there, we only have the official specs for TR and not sure even if that can be called reliable :)

Cheers
 
damn! stable 4.7GHz across 10 cores at 1.25V means 20 threads x 4.7ghz yikes
 
Are you comparing single-thread performance there? Or was those just for kicks?

Let me help you: random 1700x result: https://browser.primatelabs.com/v4/cpu/3135075

I guess this means TR is going to be only as fast as 1700x even though it has double the cores / threads. But we all know that only AMD can achieve this kind of feat.

***All of the TR results on geekbench show scores withing a couple % of each other***

Either they are ALL fake and ALL based on the same CPU, or there likely some pre-release bios issues to iron out. I am not the lord of geekbench. this is what I found, take with as much salt as you want
 
I guess this means TR is going to be only as fast as 1700x even though it has double the cores / threads. But we all know that only AMD can achieve this kind of feat.
Oh come on, don't deny that, it would be beyond hilarious if it turned out that without patching software you only get to use first 8 cores on TR.
 
Looks like voltage needs to stay under 1.3 volts or thermals get out of control and it's sucking some juice for a Intel cpu. 4.7 ghz on 10 cores is pretty good tho but I dont think the 12 core and higher will be hitting that overclock based on the thermals and wattage this chip is using.

Be aware he used a engineering sample.
 
All you guys see the temps right? 4.6ghz 1.25v and a toasty 100c in cinebench. Run something even more stressing and you are going to cook your processor to death. This is on a 240mm liquid cooling solution. So you are looking at 4.2-4.3 everyday usage with reasonable temps.
 
  • Like
Reactions: N4CR
like this
And I wouldn't be surprised if these are cherry picked chips. Wait for [H] to review some retail samples you will get the real true results with everyday usage. I doubt H will recommend running your processor at 100c under load.
 
Gotta love it fanboys only mentioning overclocks and leaving temps out. lol. Yea all the HEDT people are going to run liquid nitrogen to cool their shit. Lets get expectations in order. 4.1-4.3 max for reasonable temperatures.

You all lose reasoning picking brands. Weather its AMD or Intel.
 
Gotta love it fanboys only mentioning overclocks and leaving temps out. lol. Yea all the HEDT people are going to run liquid nitrogen to cool their shit. Lets get expectations in order. 4.1-4.3 max for reasonable temperatures.

You all lose reasoning picking brands. Weather its AMD or Intel.

What are you on about? 4.6/4.7 at 1.22v temperatures are just dandy, it's only nearing the 1.3v mark where temps get out of control
 
All you guys see the temps right? 4.6ghz 1.25v and a toasty 100c in cinebench. Run something even more stressing and you are going to cook your processor to death. This is on a 240mm liquid cooling solution. So you are looking at 4.2-4.3 everyday usage with reasonable temps.
Or you just purchase it pre-binned and pre-delidded from Silicon Lottery.
 
All you guys see the temps right? 4.6ghz 1.25v and a toasty 100c in cinebench. Run something even more stressing and you are going to cook your processor to death. This is on a 240mm liquid cooling solution. So you are looking at 4.2-4.3 everyday usage with reasonable temps.
Preschott v2.0!
Seriously though, I guess this time around we can bag Intel for being hot and room heaters for a change? That lightbulb or so of power difference on a HEDT platform, my goodness, it was such a big deal to make in the past when AMD was making hot CPUs....
 
Temperature and heat output are different things.
Technically it has both this time.
Seriously though, I guess this time around we can bag Intel for being hot and room heaters for a change? That lightbulb or so of power difference on a HEDT platform, my goodness, it was such a big deal to make in the past when AMD was making hot CPUs....
Yes, for 2 months until reviews of 4Ghz ThreadRipper hit, and we get to witness our first 400W CPU in a while.
 
Or you just purchase it pre-binned and pre-delidded from Silicon Lottery.
>Hey guys you have to buy a warranty void from get go chip if you want to OC anywhere on Intel 'HEDT' platform. No OC warranty for you!
That's a load of shit and you know it.
Noticed quite a bit of variability in OC results reported. Bet they're pulling an 'overclockers dream' with the high core parts, hence no figures.

Temperature and heat output are different things.

No shit. But with large, high end sillicon OC'd to extremes this usually means plenty of BTUs.
350 watts is the platform load mentioned in the article at 4.7Ghz - that's quite a bit of heat to dissipate. OC vs stock at least 2x power usage of a Ryzen system for for 10-20% performance? That's Presshot v2.0

So what I take from this is that I can pay 2-2.5x as much for an x299 7900 rig, which is only 10-20% faster OCd than a stock Ryzen 8c, it uses over twice the power doing this and pretty much requires water to do anything approaching 7700k levels of single thread performance? What an amazing bargain. An overclockers dream I say!
Or get a 16 core TR system for less $ which will kill it in threaded tasks and use a whole rig less power?

This is hilarious TBH.


Technically it has both this time.
Yes, for 2 months until reviews of 4Ghz ThreadRipper hit, and we get to witness our first 400W CPU in a while.

Your beloved Preschott 2.0 is pulling 350W at 4.7GHz. At 1.3V you're already over 400W - so you've already seen it. Nice to have the shoe on the other foot for once, eh? That said, Intels process scales better but holy fuck it gets hot just like Zen does.
TR @ 3.6 will be about 180-200W. You'd be an idiot to bother pushing for 4GHz (dat GHz fixation is still strong here it seems) as the 2nd critical is already at 3.6. 3.8 I would expect most to reach without huge amounts of power.
 
Haha.

I *Might* have finally found cpu's worth of the time and effort it takes to setup the Thermalright TRUE120 coppers I own.I have 3 waiting.
205 watts , yawn.
I've pushed 500 watts though one, no problem.
Intel needs to do better !

:p
 
Technically it has both this time.

Yes, for 2 months until reviews of 4Ghz ThreadRipper hit, and we get to witness our first 400W CPU in a while.

yea 6 more cores as well lol. How about intels 18 core? Yea I bet it comes at 3.2ghz base or even lower as it heats and is out of control at anything higher.

Lets just face it. No one is going to run a processor overclock it and run it at 100c under water. Not happening.
 
Or get a 16 core TR system for less $ which will kill it in threaded tasks and use a whole rig less power?

The 16C / 32T TR will cost more than the 10C / 20T i9 and also use more power.
 
It may be close to the 10C price but it will not beat it in power usage.

WTF? Now amd has to beat a 10 core intel chip in power usage? You kidding me. ROFL! Get out of here.

WTF happened to fair comparison. lol
 
Your are the one complaining about how hot the i9 10C CPU is under load. And you said above that the entire TR system would use less power than the i9.
 
When u compare the Hexus and Bittech reviews, their power consumption findings are two completely different story. Retail 7900k should be similar to 6950k at similar clocks.
 
Last edited:
350 watts is the platform load mentioned in the article at 4.7Ghz - that's quite a bit of heat to dissipate. OC vs stock at least 2x power usage of a Ryzen system for for 10-20% performance? That's Presshot v2.0

So what I take from this is that I can pay 2-2.5x as much for an x299 7900 rig, which is only 10-20% faster OCd than a stock Ryzen 8c, it uses over twice the power doing this and pretty much requires water to do anything approaching 7700k levels of single thread performance? What an amazing bargain. An overclockers dream I say!
Or get a 16 core TR system for less $ which will kill it in threaded tasks and use a whole rig less power?

This is hilarious TBH.

Indeed, it is hilarious that the more faster and efficient chip is being criticized.

53tnrr.png
 
Other than the excellent single threaded performance & additional newer instruction sets, stock i9-7900x appears to be within the same ball park performance as the haswell E5 2683 v3 with modded BIOS for much cheaper price (~$400 or less if you are patient in ebay ) in Cinebench test. Did I get this right? I'm coming mainly from multithreaded performance (e.g. in distributed computing, in which I'm actively involved), at least comparing within Intel CPUs product families. Sure the 2p xeons are not overclockable except for BCLK.

upload_2017-6-16_17-1-54.png upload_2017-6-16_17-4-28.png
 
Last edited:
>Hey guys you have to buy a warranty void from get go chip if you want to OC anywhere on Intel 'HEDT' platform. No OC warranty for you!
About time, actually. Yes, i hate cheaters and RMA on OCd chips is cheating.

Your beloved Preschott 2.0 is pulling 350W at 4.7GHz.
300W, actually, learn to substract.
TR @ 3.6 will be about 180-200W.
TR will consume more power than 2 1800Xs at same clocks. And 2 stock 1800Xs consume over 200W already.
 
And I wouldn't be surprised if these are cherry picked chips. Wait for [H] to review some retail samples you will get the real true results with everyday usage. I doubt H will recommend running your processor at 100c under load.

Need to consider like I mentioned before we have no idea where the CPU and motherboard came from for both the Hexus and Bit-tech review because they are outside the NDA, meaning they borrowed these from some source and not necessarily good examples.
Case in point Hexus has a different result to Bit-Tech and hitting 100c when going to 1.3V rather than lower voltages like Bit-tech
In fact 1.25V in the Hexus review managed 4.7GHz all cores with a Noctua D15S, they then swapped it out to do higher voltages towards 1.3V with AIO.
Never know could be something to do with the motherboard/BIOS and pushing too much voltage in the Bit-tech review (happened in the past), Hexus results seem to align more with Der8auer and what was hinted by 8pack at overclockersuk.
But yeah really need reviews where everything is using current HW/BIOS/etc, for loaners and reviews outside of the NDA not sure that can be fully relied upon.
Cheers
 
Back
Top