Intel Core i7 7820X Skylake X Review

It already works. Ryzen is made in quad core clusters.
430b7ed75898.jpg


I see a difference, and the possibility for problems.
 
430b7ed75898.jpg


I see a difference, and the possibility for problems.

Infinity Fabric is what they are using. And it's not a problem. Unlike using more than one CPU the latency is almost 0 and it does not present itself as multiple CPUs.

If you buy this for its intended use you will have no problems. Most of those who will buy 16+ cored CPUs have the wherewithal and capacity to code their own software for it any way.

If you buy it to wave your gherkin around it's likely you will be disappointed.
 
i am only interested how threadripper works for 3d max, maya, after effects and premiere pro.

could not care less about playing games on TR....
 
compared to the latest intel x299 offerings....

Take Skylake IPC @ comparable clock speed.

Take Ryzen IPC @ comparable clock speed.

Example - 8c 16t Ryzen scores 1400 points in Cinebench @ 3.2ghz. It already seems to work with Infinity Fabric, so for the 16c 32t Ryzen at the same clock speed you can expect around 2800+ points in Cinebench.

TBH after doing some calculations earlier Ryzen seems to stand up really well against the 7820x. On IPC terms per clock any way. The *only* thing this $600 or £600 has that the AMD one doesn't is clock speed. Because apples for apples mhz to mhz there is very little in it.

AMD are kicking Intel where it hurts them the most. Cores and PCIE lanes. Intel love derping cores, lanes and locking vs unlocked. AMD are going to kick them where it hurts. Loads of cores, loads of lanes, unlocked.

So Intel are following suit now, but are double the price of the AMD for what? 500mhz.

I have read other reviews of the 7820x and they were stuck at around 4.5ghz with crazy temps.

Thing is as we all know once you hit around 4.5ghz on any modern CPU gaming performance slopes off. So do you need any more than that? simple answer is no. For years and years we have been insisting that our GPUs do all of the hard work so that is what we have. CPU performance is not anywhere near as important as people seem to think. You are still better off with a entry level CPU and high end GPU, for example.
 
yeah thats all fine speculating. i can do that too. :)

but the real world tests will show how much this is worth.

i never expected the skylake-x 10 core to have such bad thermal behavior.
i thought it would be on par with the broadwell-e.

everything else is basically close to what i thought.

maybe TR has some good/bad surprises too.
 
yeah thats all fine speculating. i can do that too. :)

but the real world tests will show how much this is worth.

i never expected the skylake-x 10 core to have such bad thermal behavior.
i thought it would be on par with the broadwell-e.

everything else is basically close to what i thought.

maybe TR has some good/bad surprises too.

A little bird in Taiwan says that Intel are working on soldering later CPUs..

Broadwell E are soldered. That is why the disparity.
 
If you're referring to that obscure bitsnchips website, please don't :) They may be soldering 12-18C parts, that I'd believe, but not the current offerings.

Nope apparently they are working on soldering all of them.

Please don't believe what they have told you. They don't give a f**k about you or your overclocking ambitions. They are spitting in your face. In the same way they spat in every enthusiast's face when overclocking started. They immediately did everything they could to stop it, and any "accident" they have left in place has been shut down pretty quickly.

The only reason they are finally figuring out what to do about it is AMD. Before that they couldn't have given a rat's arse.

It's kinda like having a friend that does nothing but dump on you and take your money, then some one comes along to compete for your friendship and your crap mate finally starts treating you better. Not because he wanted to, because he has no choice.

If you don't think a billion dollar corporation like Intel can work out a way to solder CPUs? your lookout dude. AMD have it down pat.
 
Nope apparently they are working on soldering all of them.

Please don't believe what they have told you. They don't give a f**k about you or your overclocking ambitions. They are spitting in your face. In the same way they spat in every enthusiast's face when overclocking started. They immediately did everything they could to stop it, and any "accident" they have left in place has been shut down pretty quickly.

The only reason they are finally figuring out what to do about it is AMD. Before that they couldn't have given a rat's arse.

It's kinda like having a friend that does nothing but dump on you and take your money, then some one comes along to compete for your friendship and your crap mate finally starts treating you better. Not because he wanted to, because he has no choice.

If you don't think a billion dollar corporation like Intel can work out a way to solder CPUs? your lookout dude. AMD have it down pat.

Chill a bit for your own sake? :) You've conveyed your dislike for Intel and enthusiasm for AMD well over and over again regarding these matters, it's all good :cool: We can only guess at why Intel used TIM, regardless of how bad we'd like a straight answer out of them. Reminder: if I'm not mistaking Xeon Phi uses the exact same TIM instead of solder? And those are e x p e n s i v e ! Translation: there could be other reasons. Intel doesn't care about me. AMD doesn't care about me. G.Skill doesn't know my name either, and I'm pretty sure the birds in my garden have forgotten I exist. It's business mate, and I care as much about them as they about me; I care about what I can get to do what I want and need. Long story short; if they remake all with solder, great!

Dambusters-The-Opening-Shots-lg.jpg


THIS. Barnes Wallis would expect no less.

One of my favourite books as a kid, and recently I went to that lake in the woods where they did their practice rounds, kinda nostalgic.

No bombs dropped yet then? I got a tip that maybe I misinterpreted Tom's reply to hold off; he may have been referring to holding off on the 7820X in light of some Coffee Lake-S information he has, instead of holding off on CFL-S?
 
Chill a bit for your own sake? :) You've conveyed your dislike for Intel and enthusiasm for AMD well over and over again regarding these matters, it's all good :cool:

Perfectly chilled, thanks.

I have not conveyed any such thing. Go and read the Vega thread, for example. I am just calling it as it is which is what I always do.

AMD Ryzen - amazing.
Intel X299 - more contempt.
AMD Vega - complete horror show. And so on.
 
AlienALX: there IS latency between the CCXs. This is well known, so your "zero latency" statement is patently false. Why do you think tech forums everywhere are abuzz with how to get RAM speeds higher with Ryzen? It's because of the inherent latency with the fabric, and that improves with higher rams speeds. Unfortunately Ryzen isn't anywhere near as amazing as you make it sound. Yes, it's a great bang for the buck, but it takes a LOT of tweaking to make it run fast, and it's unstable as hell right now, even in STOCK form. It sure as hell isn't plug and play. Unless they iron out these issues, Threadripper will be a massive fail.

If you bought a Ryzen rig in the past 2 months, you unwittingly signed up to be a beta BIOS tester, and you aren't getting paid for it either. :)
 
This I agree with 100% but not just bios - youre an active hardware tester.

Agreed! Ryzen was a SUPER rushed release, and the kinks are still there months later. I wonder how many more months it will take for these AGESA updates to improve, and if that will even solve most peoples issues.

I've flashed my C6H BIOS about 15 times with multiple RC BIOSes, and even the mighty Asus isn't close to getting the kinks worked out yet. It seems with every BIOS, some things get better, others worse. This latest RC BIOS (1401) seemed to improve memory compatibility, but broke my RGB LEDs (they work, but they don't shut off on power down, even though the BIOS says to). That *used* to work fine, but broke in the latest BIOS. But not for everyone, just me, go figure. Asus is scrambling in the wake of all the Intel stuff, and it seems like things are slipping through the cracks.

There's a thread on OCN that is WELL over 2000 pages, and it's FILLED with stuff like this.
 
Agreed! Ryzen was a SUPER rushed release, and the kinks are still there months later. I wonder how many more months it will take for these AGESA updates to improve, and if that will even solve most peoples issues.

I've flashed my C6H BIOS about 15 times with multiple RC BIOSes, and even the mighty Asus isn't close to getting the kinks worked out yet. It seems with every BIOS, some things get better, others worse. This latest RC BIOS (1401) seemed to improve memory compatibility, but broke my RGB LEDs (they work, but they don't shut off on power down, even though the BIOS says to). That *used* to work fine, but broke in the latest BIOS. But not for everyone, just me, go figure. Asus is scrambling in the wake of all the Intel stuff, and it seems like things are slipping through the cracks.

There's a thread on OCN that is WELL over 2000 pages, and it's FILLED with stuff like this.

The average consumer who is struggling to get his Hynix memory to run at 2933Mhz on a particular BIOS that seems to allow him to overclock his 1600 to 3.95Ghz instead of 3.85Ghz is where it matters the most for the average consumer. OCN has always been a huge support system for people in those situations. It's where I always go for the widest array of opinions and experiences.
 
That's the one caveat at being stuck at lower RAM speeds. Your OC will likely work a LOT better. My R7-1700 is a dud unfortunately it seems. Won't even be stable at 3.8Ghz without insane volts. Yet at stock 3-3.2ghz, it just sips voltage, run cool as a cucumber, etc. But the jump to 3.7+ requires such a huge voltage jump that I'm not sure it's worth it.
 
Back
Top