UglyDuckling
Senior member
- May 6, 2015
- 390
- 35
- 61
Which is a truly incredible statement when you are talking about 4c/8t versus 8c/16t. This bodes extremely well for Raven Ridge mobile parts.I would call the 1700X, 7700K, and 1800X a tie for power consumption.
Idle power is high, it's a lot higher when overclocking your RAM, and it's a little odd/concerning that per-core power drops when OC'ing your RAM.well, the results are basically the expected, I think it's positive that the performance hit even for gaming due to the "glue" is not all that significant on those early tests, also power usage is under control (given 2x R7s at high clock in there), idle power usage being a little higher
might also be a question of the MB being used
on the MT tests it clearly can shine very often, still in some things the 7900X is quite close, HEVC handbrake gets my attention, is somewhat unfortunate for TR that is so close, but I guess that's the advantage from AVX?
also interesting that a few softwares kind of take a hit I think for having to many threads, like o Dolphin the 1950X was much slower than the 1920X when looking at both in creators mode on the Anandtech test...
Which is a truly incredible statement when you are talking about 4c/8t versus 8c/16t. This bodes extremely well for Raven Ridge mobile parts.
What is with the jump to the conclusion that I was attacking AMD? The three chips were tied in power consumption. Most likely within the limits of error of measuring power consumption.That's just factual info.Which is a truly incredible statement when you are talking about 4c/8t versus 8c/16t. This bodes extremely well for Raven Ridge mobile parts.
Actually, you could probably turn that statement back around on yourself for even asserting that. I was literally saying in other words "yeah exactly, and its incredible that they are so close given one is 4c and one is 8c". I promise you I was not argumentative in my tone! Sorry for the miscommunication.What is with the jump to the conclusion that I was attacking AMD? The three chips were tied in power consumption. Most likely within the limits of error of measuring power consumption.That's just factual info.
This board is getting really tiresome with the knee jerk reactions and assumptions.
Okay, my bad as well.Actually, you could probably turn that statement back around on yourself for even asserting that. I was literally saying in other words "yeah exactly, and its incredible that they are so close given one is 4c and one is 8c". I promise you I was not argumentative in my tone! Sorry for the miscommunication.
This is what I get (power wise) at 3700MHz / 1.150V.
These figures are DCR, read directly from the VRM controllers.
Output values are what the actual domains on the CPU draw from the VRM (and therefore are dissipated by the CPU), input values are what the VRMs draw from the EPS12V connectors.
Am I reading your numbers correctly -- the VRMs are dissipating 40-60W? If so, that's a lot more than I had expected. What memory clocks are you running at?
The average IPC difference in non-AVX workloads is like ~10% which is basically not much. Clock difference @ default settings is however larger as intel parts do achieve higher all core and single core /low core Turbo clocks. For example 7900X gets 4Ghz all core Turbo for all 10 cores while 1950X gets 3.5Ghz. TR can OC to 4-4.1Ghz from what I have seen in these early reviews while intel 10 core part can get to ~4.4-4.5Ghz with "safe" 24/7 voltage. Downside is load power draw on SKL-X which even @ stock settings is much higher than on TR as per THG's review, meaning a 10C 7900X @ 4.5Ghz consuming the same amount of power as 16C TR @ 3.9Ghz when fully stressed across all cores.So in summary Intel is better at IPC at all price ranges, while AMD is better at multi threaded performance at all price ranges.
Good to see AMD making a strong comeback, we will all benefit as consumers from having proper competition again.
I've tested the older Liqtech models and based on the experience they are superior to Asetek ones.
Better structure, significantly better radiators and better all-over build quality.
The situation with Threadripper compatible coolers is extremely pathetic at the moment. There is a single cooling solution (20+ variants, all made by Asetek) available for it.
Air coolers exist from Arctic, Noctua and Coolermaster but most of them are insufficient (Arctic, Coolermaster and the smaller Noctuas) and none of them are actually available. EKWB was bragging that their Threadripper blocks were ready already in december, yet they have announced that the shipments will being on the 18th.
The gaming performance is quite surprising too, given the shrill voices of the naysayers when it comes to NUMA and games.
It is good for what it is, one 8-core CPU vs two 8-core NUMA nodes. Frankly, I expected much worse.Thats is not what i whould call "good" for a system that will cost you about 3 times as much as a 1800X system. And as games become more MT i whould expect the 1800X to become better and the 1950X to get worse as the overhead can only become worse and worse.
It's fair to presume that informed Threadripper buyers will view gaming performance as a secondary consideration, as such, the fact that it does fairly well is a plus.Thats is not what i whould call "good" for a system that will cost you about 3 times as much as a 1800X system. And as games become more MT i whould expect the 1800X to become better and the 1950X to get worse as the overhead can only become worse and worse.
It is good for what it is, one 8-core CPU vs two 8-core NUMA nodes. Frankly, I expected much worse.
That is a weird statement to make.Me too, but it is still bad, and chances are it will get worse.
It's fair to presume that informed Threadripper buyers will view gaming performance as a secondary consideration, as such, the fact that it does fairly well is a plus.
Yeah this point is getting more and more annoying. It's barely a thing on R7. But anyone looking into high core count systems is buying very very expensive components in comparison. They also know that even on Intel due to clock speeds that the higher the cores the worse the gaming performance on older games. So the older games that people are trying to run at 144+ FPS aren't going to do as great as a 7700k.Not to mention for those of us for whom gaming is at best a secondary consideration, we're not gaming at 1080p high settings...
That is a weird statement to make.
Not to mention for those of us for whom gaming is at best a secondary consideration, we're not gaming at 1080p high settings...