Der8auer, a well known professional over-clocker, has put the hex core Ryzen 5 under liquid nitrogen. It started beating previous records held by the 5820K (Haswell).
It is surprising that it is beating the 5820K even though it is clocked lower. The 5.4GHz/5.5GHz clocks shown in the video below have the RAM running at 3100 MHz, the 5.9 GHz clock does not show what the frequency of the RAM is.
Note that the 5.9 GHz clock only registers 0.4V on the core (CPU-Z). Going on the 1.9V+ needed to get the CPU to run at 5.4GHz, I can only guess that they easily broke the 2.0V mark. He does not reveal why CPU-Z is only showing 0.4V in the video. If anyone has insight into this, please share.
My best guess is that they had trouble with getting the hardware to accept the high voltage settings. What Der8auer has done before is "trick" the hardware into believing that less voltage is being requested, less than the true current being thrown at the hardware. I suppose it could be done with a custom BIOS, or de-soldering resistors on the motherboard or something.
This was a nice little surprise, but don't let this fool you. Most members here will immediately understand that these are not viable clocks for most of the Ryzen user base. Seems, for this generation anyways, we will have to settle for 3.9 GHz - 4.2GHz on air/water in our PCs.
You know it's bad for Intel when an AMD system is topping the benchmark records. This a thing that Nvidia has had a monopoly on in the Firestrike benchmarks that keeps them in the mindshare in the GPU market, if this keeps up AMD's mindshare might just grow.
The 1700 is only $100 more. That would mean 25% more multi-thread performance. If you are "future-proofing" or into video/photo editing, that is not a bad investment... if you are not on a tight budget.
I don't run my browser like that but I would like to think that having tabs open in the background would not demand much CPU performance. You do have a good point though.
Ultimately, I was just curious as to what kind of customers flazza caters to.
On Newegg it's 249 for the 1600X vs 327 for the 1700. That's only 78$. But the 1700 is 27% more than the 1600X for your 25% more multi-threading. Now put non-linear scaling into the mix and it does not seem quite so clear that the 17001600X is the best bang for the buck.
Plus: 249$ is under 300$ and 329$ is over 300$. That makes a psychological difference. (Why else would everything cost 299.99$ instead of 300$?).
I also like Vivaldi's Tab hibernate feature (might be a Chrome/-ium stock feature, not sure), but sadly there doesn't seem to be an option for automatic hibernation (yet).
Why else would everything cost 299.99$ instead of 300$?
Just friends and family who are far less technical, they have all bought such lemons in the past that all builds now go through me.
The stories I could tell...
Like my sister buying an all in one with integrated graphics and no expansion ports for her son and getting annoyed that it cant play games.
Or my mum getting an xps laptop when all she does is use it for facebook and youtube.. who strangely after I bought her an android tablet has never touched that laptop ever..
Or my brother buying a 650 ti at crazy shop price (for the time) when he could have bought MUCH better for the same price online (like two tiers higher).
My family and friends just cant seem to be trusted to build / buy sensible pc's
Might be a while until I get another rebuild as all the pc's I have built for them have performed like champs.
As I said though, as soon as they get to the end of their life my choice will be 1600, b350 and a 470 (or nvidia equivalent, all comes down to price / availability).
Would be great though if AMD just lay off the Polaris crap already. I want some Vega and HBM2 stuff. I can't believe AMD not only caught up with Intel CPUs (or the mainline ones, not the Xeons yet), but made them a better option, I could easily go for Ryzen 5 over a Core i5 and a Ryzen 7 over the quad core or six-core i7s. But I have no reason to upgrade the CPU since my i7 5820K does just fine.
But the Ryzen 5 1600 is $220, so that would actually be just as much of a good value as the Ryzen 7 1700.
Also, when it scales non-linear, it's diminishing returns, going from 6 Cores to 8 Cores isn't as much of a benefit as going from 4 Cores to 6 Cores and nowhere near 2 Cores to 4 Cores.
I definitely intrigue in what VEGA can do with a proper driver. AdoreTV says the DOOM demo at CES is the best case scenario for it and doubt drivers official drivers would improve it any further, but we'll see.
If you just count cores the 1600 might be the better value. If you take into account that there will be always that legacy-single-threaded piece of code that has to be run (Looking at you Photoshop, Arma 3 and whatever else your name is).
Yes you are right on the non-linear scaling. But this is a multi-dimensional optimization problem. You can optimize: cores / $ base-clock / $ boost-clock / $ cache / $ cache / core / $ or any number of properties.
So do you want to tell me that one should not considerate the price per additional core for a CPU because the anything above the dual-core is costing more than it delivers per core? If one would only go by that metric that would be the case. But you and I know that you consider more than one metric to conclude which CPU would be better.
Intel should definitely be scared that this chip (a chip that is literally a badly binned chip, or just has cores disabled) is beating their fully unlocked and fully working chips. I'm excited for Ryzen 3 because I think it will be perfect for a low-power NAS or PfSense box.
Well one important detail i´m missing is the core voltage at that OC. I mean most people dont seem to be able to overclock their Ryzen chips over 4.1Ghz under normal conditions. This seems to be some limmitation with those chips or something. So i´m pretty currious how they got arround it. Custom firmware or microcode or something maybe? Or might that 20°C offset temperature has anything to do with Ryzen´s overclock limmitation maybe?
I have seen these particular guys do hardware mods to "trick" the system, and I think a firmware mod would be capable of this as well... in some situations.
Maybe they are keeping what they did a secret for now... that way they can hold their record scores for a little while. They usually do show their work though, it bothered me a little.
I also wanted to know if there was any silicon degradation due to all that voltage being pumped through the chip for extended periods of time.
