AMD Threadripper Is Looking Good, At Least Under Windows

[Luke_Wolf]

If that's true, then it'd be really great news, as all those who encounter the segfault issue might be able to get it resolved by getting a replacement CPU. Given the fact that some people already received replacement chips which didn't segfault, probably the explanation is somehow correct.

bridgman can we get a confirmation on this? I don't want to RMA if I don't have to.

[duby229]

bridgman can we get a confirmation on this? I don't want to RMA if I don't have to.

Why wouldn't you? If you have an affected chip, then an RMA is your only option. You can alleviate the problem so much so that you won't experience it, but at the cost of disabling features and functionality. The only fix is RMA for a known good one.

[hoohoo]

I appreciate the work Michael does in benchmarking overall performance, but he doesn't scratch the surface when it comes to understanding why the performance has changed. <...>

Certainly anandtech does more in depth reporting of vendor technical slides when it comes to details of a CPU or GPU arch. But really, what Anandtech presents for benchies is mostly code running on Windows. I think pcperspectives has done better work than anandtech for some time. For example pcper was the first (AFAIK) to demonstrate the difference between intra CCX and inter CCX latency on Ryzen. Regarding THG, I can't comment because I don't read it much.

I generally want to know how GCC/G++, Clang, or the kernel runs on a new CPU. Michael gives me that and I give Michael advertising clicks. I'm happy.

Nobody that I know of is running proper compute oriented GPU benchmarks, just games.

Michael has a test suite he runs, just like all the Windows hardware/gaming sites have their suites. Perhaps you would be happier if Michael would pump out the verbiage like anandtech does?

What is Ian Cutress's favorite irrelevant word the past few quarters? "Poignant"! Benchmarks results have deep emotional weight for him it seems. I'll take "just the facts, Jack", for Michael's Linux results. I'm OK with that.

[hoohoo]

... Michael needs an editor. ...

Your arrogance in that comment is beyond belief.

Go start your own Linux gaming & hardware news & reviews site, run it for a decade or so and then report back when you've walked in his shoes.

[linuxgeex]

Your arrogance in that comment is beyond belief.

Go start your own Linux gaming & hardware news & reviews site, run it for a decade or so and then report back when you've walked in his shoes.

It warms my heart when I see people doing something they believe in support of something they believe, so kudos for that, I guess.

[linuxgeex]

And if he can't find more interesting topics to include, then what? Advertising revenue and memberships drive the site, and no matter how much we hate it, his habit of posting five or so articles a day does draw more page views than if he only put big news items up once in a while.

i.e. He makes mountains out of molehills and flirts with clickbait headlines because it pays the bills. If he had another ten thousand paying subscribers, that would change and the site could get more serious. But if that was going to happen, it would have happened already.

I think you're right about ad revenue. I think you're wrong about subscription / donation revenue. The ratio of the two probably says something about that, but I'm not privy...

Anyhow, I'm glad not everyone has the opinion that reducing quality is the best way to make more money. If so, the only restaurant would be McDonalds, lol.

[linuxgeex]

I think that has been tried already:

http://dilbert.com/strip/1996-04-15

True enough, I've been on both sides of that situation over the years... having management press me for documentation when I was behind on production (resulting in me having to make wiser use of my time to meet the new expectations), and also putting similar pressure on someone to force them to make wiser use of their time, and it worked there too. So that tactic isn't always as asinine as it first appears.

[bridgman]

The problem I have with that whole "bring in another layer of management" thing is that it only works if everyone was "working stupid" before. After the 10th or 15th "let's bring in more management" cycle it starts to seem unlikely that there is any more stupid left in the system to resolve.

[AndyChow]

Wow, a lot of hard comments. I think Michael needs two things: Time off and part-time employees.

IMO, he works much too hard, and the few times he has taken a day or two off, he still cheated and posted several articles during that time.

Also, he needs to find some part time employees that will submit articles and research to him.

He's running a one-man empire. Literally. This site, the test-suite, the automation framework, and so on.

Michael, it's okay to reach out and try to delegate a bit. I personally enjoy much of what you do, but I don't want you to burn out from the pressure.

I think you like Oktoberfest, maybe try to set up a few people to write and submit articles for you during this period. You get final approval and the power to post it or not, but at least take some time off and enjoy a bit.

[ssokolow]

Graphics aren't the problem for UT99, modern CPUs are. For one, frequency scaling - so the game doesn't constantly go through speed up/slow down cycles, you need to use a fixed frequency CPU governor. The other problem is the game running super fast (like really super fast), this is hacked around by having the launch script stress the CPU 100% while the game launches. This is for the Linux version I have installed, no clue what's going on in the Windows world, though a quick googling shows me there's a D3D11 renderer for the game nowadays!

Trust me, I'm aware of the CPU assumptions made by the original Unreal engine. I think the rendering switch was to deal with some kind of weird driver incompatibility.