AMD - Zen chitchat

[kompukare]

Guess that all depends on the quality and longevity of the fans, since generally most miners underclock. Back when it was Litecoin and Hawaii was the go-to card, people managed to underclock and undervolt a 250W card to 120W or so. Or that was certainly the case for anyone whose electricity wasn't 10c or less, or who also had to pay for air conditioning.

So if anything (compared to heavy gamers), the rest of the card shouldn't have gotten much usage, just the fans.

[watercooled]

Is their any evidence of mining damaging cards? I can't see why they'd be half broken.

Yes, even the AIBs complain of exceptionally high RMA rates for cards used for mining.

Guess that all depends on the quality and longevity of the fans, since generally most miners underclock. Back when it was Litecoin and Hawaii was the go-to card, people managed to underclock and undervolt a 250W card to 120W or so. Or that was certainly the case for anyone whose electricity wasn't 10c or less, or who also had to pay for air conditioning.

So if anything (compared to heavy gamers), the rest of the card shouldn't have gotten much usage, just the fans.

Fans are one thing - back around 2013 I remember there were some fans failing due to leaking bearing lubricant as they were being set to 100% 24/7 which they were never designed for. Bearings have a rated lifetime which obviously depends on speed, and the stock fan profiles usually wouldn't have had them close to 100% even for short periods in a gaming PC.

I think there are failures related to VRMs and VRAM too, but if I had to guess, that's likely down to inadequate cooling (e.g. by jamming a load of cards back to back on a rack and not keeping them cool). Basically it depends on how they're used (or abused) AFAIK - if they're kept sufficiently cool, not overclocked/volted or had power limiters increased, it probably wouldn't cause all that much harm to well-engineered cards. But as scaryjim says, length of use does factor in to reliability.

[Xlucine]

I wonder how hard it is to buy a replacement fan motor from the manufacturer? I suspect it isn't easy

[watercooled]

The broken fan itself isn't necessarily what I'd be most concerned about - even if the whole card was adequately cooled with the fan - how long was it running with impaired cooling? Dust is another thing to consider.

If it's been bumping up against the throttle point for extended periods of time, chances are the VRM and memory will have been extremely hot too. That sort of thing does kill cards.

[kompukare]

Never mind the performance of Threadripper, I found something far more interesting to do:
tally up all the 250 press Threadripper review kits with the reviewers!
So put them in a spreadsheet hosted on ethercalc.org:


https://ethercalc.org/njqz3ceikayq

Hm, only have ten so far. Most of them are youtubers as 'Threadripper unboxing' videos were east to find. And that's probably as far as I have patience for watching youtube reviewers.

However, that Ethercalc spreadsheet is open for anyone to edit...

[Corky34]

AMD has confirmed Ryzen's got a problem under certain conditions on Linux.

From what they say it seems under certain conditions, heavily loading multiple threads along with less than optimal cooling, causes Ryzen to read/write too the wrong memory location, i guess because of a localised hot spot within the silicon.

[CAT-THE-FIFTH]

AMD has confirmed Ryzen's got a problem under certain conditions on Linux.

From what they say it seems under certain conditions, heavily loading multiple threads along with less than optimal cooling, causes Ryzen to read/write too the wrong memory location.

Interestingly enough Epyc and Threadripper don't have the same problem so it makes me wonder if they are using a different stepping.

[Corky34]

They also say "early Ryzen processors" so maybe a manufacturing issue that's already been addressed?

[Bagnaj97]

Interestingly enough Epyc and Threadripper don't have the same problem so it makes me wonder if they are using a different stepping.

Could be differences in microcode? If so it's hopefully an easy fix.

[CAT-THE-FIFTH]

First Techspot/Hardware Unboxed and now DF,highlight how great a CPU the Ryzen 5 1600 is:

http://forums.hexus.net/pc-hardware-...me-battle.html

Its the new Core i5 2500K!!

[watercooled]

AMD has confirmed Ryzen's got a problem under certain conditions on Linux.

From what they say it seems under certain conditions, heavily loading multiple threads along with less than optimal cooling, causes Ryzen to read/write too the wrong memory location, i guess because of a localised hot spot within the silicon.

I've been following this issue for a while - it's hard to gauge from a forum thread how common an issue actually is, so it's good AMD have (finally) recognised it. It's still really not clear what's causing it - it's likely nothing architectural or there wouldn't be any fault-free, and AFAIK there's only the one stepping. It could be something to do with power delivery i.e. brownouts occurring in very specific parts of the die under certain scenarios? In theory that should still be fixable with a microcode/firmware patch to get the power microcontroller to better compensate for it.

It's interesting nonetheless, but it's good it's been weeded out before the workstation/server chips hit the market!

On a similar subject - I'm still waiting for MSI to release a BIOS update for that Skylake bug!!! Asus had one out ages ago.

[CAT-THE-FIFTH]

I am surprised not more of you have commented on the DF article and some of their findings. It makes me wonder whether we need to be looking in more detail at what reviews are looking at!!

[Corky34]

It makes me wonder whether we need to be looking in more detail at what reviews are looking at!!

Such as? I'm not asking that in a belligerent manner it's just don't their findings tell us what we already knew, that overall a higher clocked Intel chip is going to perform slightly better in games and if monies no object then you're probably going to buy an Intel chip, if on the other hand you want something more than an expensive glorified gaming console then you'd probably be better off with Ryzen.

Disclaimer: That's just my opinion so please don't flame me or ask me to go digging around the Internet in an attempt to validate it.

[CAT-THE-FIFTH]

Such as? I'm not asking that in a belligerent manner it's just don't their findings tell us what we already knew, that overall a higher clocked Intel chip is going to perform slightly better in games and if monies no object then you're probably going to buy an Intel chip, if on the other hand you want something more than an expensive glorified gaming console then you'd probably be better off with Ryzen.

Disclaimer: That's just my opinion so please don't flame me or ask me to go digging around the Internet in an attempt to validate it.

Watch it as it's very interesting. The Core i5 CPUs were producing more framerates in places where nothing was going on bumping up their averages where the Ryzen 5 was showing more framerates in areas which were CPU heavy.
Hence unless we see the framerate graphs or a video looking at averages or even the minimums on a chart might not be a good idea.

Edit!

Even at 4.8GHZ the Core i5 really couldn't really best the Ryzen 5 1600 when it came to minimums in a few games.

[Corky34]

I did watch it and it just reinforced my opinion that if all you care about is maximum FPS then Intel is for you, if on the other hand you want a better all round balanced system then Ryzen is for you, yes a single set of benchmarks often don't reflect all the strengths and weaknesses of particular hardware, even the DF article you posted doesn't, however, speaking personally, when buying hardware i review as much of the available information as i can, but that's because I'm time rich and money poor.

I would guess the majority of people just want to be told what's the best thing they can buy for their money and most review sites define "best" to mean fastest.

[CAT-THE-FIFTH]

I did watch it and it just reinforced my opinion that if all you care about is maximum FPS then Intel is for you, if on the other hand you want a better all round balanced system then Ryzen is for you, yes a single set of benchmarks often don't reflect all the strengths and weaknesses of particular hardware, even the DF article you posted doesn't, however, speaking personally, when buying hardware i review as much of the available information as i can, but that's because I'm time rich and money poor.

I would guess the majority of people just want to be told what's the best thing they can buy for their money and most review sites define "best" to mean fastest.

So the "maximum FPS" for Intel is just trying to make the Core i5 7600K look better,when DF have said as a gaming CPU the Ryzen 5 1600 is better. If you care about maximum FPS you would not be getting a Core i5 7600K since it does not produce as high minimums in newer games,and you don't seem to get that many websites just put on an average FPS chart,which means the Core i5 scores are overinflated.

So that means,the RL gaming experience is going to be worse.

Its going to get worse and worse over the next two to three years too,but by then the 6C Core i5s will be out and that will be quietly buried.

The Witcher 3, Rise of the Tomb Raider and especially Crysis 3 love frequency of course, but they also benefit heavily from as many cores and threads as you can throw at them, with both Ryzen 5s offering a significant advantage over Intel's stock Core i5. Here, Ryzen 5 sits comfortably at a mid-point between Core i5 and Core i7. However, not everything is as it seems based on the numbers alone. Assassin's Creed Unity posts a lead on the i5, but when studying performance at the per-scene level, i5 is pulling ahead in relatively empty scenes in our benchmark, with Ryzen 5 performing better in areas packed with NPCs. There's the suggestion that the i5 frame-rate average is boosted by big performance gains in less useful, more 'empty' rendering scenes. Similarly, in the Crysis 3 benchmark, the i5 and indeed i7's scores are skewed higher when the viewpoint shifts to similarly sparse scenes.

We've got a complete breakdown of this behaviour in our video review, but the bottom line is that it's not just different game engines that can favour i5 or Ryzen 5 processors - it can actually vary on a scene-by-scene nature in many games. And of course, therefore, results can vary depending very much on what scenes are chosen for benchmarking.

To illustrate, The Witcher 3's Novigrad City - our test area - can easily max an i5 quad with 100 per cent utilisation across all cores, and Ryzen 5 is faster here. However, benchmark a less demanding area or an engine-driven cut-scene and the i5 takes the lead. We've tried to tailor our tests to concentrate more on these heavier workloads and for our money Ryzen 5 is the more versatile, capable performer in areas where the CPU matters most in gaming.

People have been so used to hearing that Core i5 is the bestest gaming CPU for the average gamer in the world,they are in shock when it apparently is not as good as they think it is!!

LMAO,DF even mocked the Core i5 in the video- the Ryzen CPU was overclocked using the stock cooler,and the Core i5 7600K and the H110i GT setup was £300+ and had a 1GHZ advantage.

This is why the new Core i5 is a 6C CPU,since even Intel thinks 4C/4T is only worthy for competing with a quad core Ryzen CPU.

Intel now thinks 4C/4T is worthy of Core i3 status.

I still LOLed at people spending £240 on a Core i5 7600K earlier in the year.

What a joke of a CPU in 2017.

Edit!!

Also remember DF is part of a massive gaming website called Eurogamer,so they actually have a much better understanding of games than most hardware review sites.