Computer Forum Chat Thread

[Remilia]

My F8500 Plasma has a green column in the middle which is a sign of a problem in an IC inside a ribbon cable, I honestly have no hope whatsoever of them fixing it since they would probably want to replace the entire panel and there are no panels available.

Rip. Yeah, that's not something you replace easily.

@Remilia CLED when

Currently only used for large format displays and not for anything consumer. So for consumers, never.
I'll just wait for CLED or the Panasonic IPS for any new screens.

[Artorius]

Rip. Yeah, that's not something you replace easily.

http://www.avsforum.com/forum/167-pl...ixels-why.html

This exact problem, someone actually managed to fix it simply reseating the cables. I could try doing this at some point.

Also yeah, with CLED being closer to us than ever and the magical IPS being a thing, there has never been a better time to just wait... We'll also get Samsung TVs with Freesync next year running at UHD@4:4:4@120 which is finally possible without multiple-inputs due to HDMI 2.1.

[Remilia]

http://www.avsforum.com/forum/167-pl...ixels-why.html

This exact problem, someone actually managed to fix it simply reseating the cables. I could try doing this at some point.

Also yeah, with CLED being closer to us than ever and the magical IPS being a thing, there has never been a better time to just wait... We'll also get Samsung TVs with Freesync next year running at UHD@4:4:4@120 which is finally possible without multiple-inputs due to HDMI 2.1.

Might as well give it a shot.

[DeltrusDisc]

Magical IPS?

[Remilia]

Magical IPS?

It's what's in Eizo CG3145 reference monitor. Basically it's a two layered LCD panel boasting 1000cd/m2 peak lighting and 1,000,000:1 static contrast ratio. Eats more power relatively speaking because it needs to pass light through two layers as opposed to one. The first layer are black / white cells that can control black levels in a more fine grained way than compared to the RGB sub pixels in the well, RGB layer. This way it fixes one of the biggest weakness of an LCD, the contrast. The remaining weakness really is the pixel transition and the motion blur. CRTs still have that on lock down.

[Artorius]

I'm impressed by Zen+ SKUs, the leap over the original Zen SKUs is enormous and they're mostly beating Intel across the board, including games which were previously Zen's weak point.

This is huge.

[Remilia]

Yeah, the decrease in memory latency really helps.

[BicycleMafioso]

I am still annoyed by how people kept and keep saying zen was 'weak' in games. They never were. The other side were stronger, but Zen was/is not weak. Bulldozer was weak in games. Zen is just "not literally the strongest".
Being second in a two man race, coming 2 seconds after the winner is still very close, compared to six minutes. (numbers arbitrary and to prove a point and not to be taken as relative performance numbers)

[Artorius]

It was never horrible, but it always the only thing you could actually call a weak point. Everything else was more or less great.

Now literally everything is great, incredibly solid performance everywhere.

[Alindra]

My monitor is barely 3 years old and it sounds like Samsung won't send me a replacement power brick. I need to buy a new, OFF BRAND one for $45. Fuck.

What cheap shits. I paid $1,200 USD after taxes for this thing. Wow.

Wow, that's some shit right there. As I recall, you were really happy with that monitor, too. Would you still recommend Samsung monitors after this?

I am still annoyed by how people kept and keep saying zen was 'weak' in games. They never were. The other side were stronger, but Zen was/is not weak. Bulldozer was weak in games. Zen is just "not literally the strongest".
Being second in a two man race, coming 2 seconds after the winner is still very close, compared to six minutes. (numbers arbitrary and to prove a point and not to be taken as relative performance numbers)

So what you're saying is you like inferior products, huh? /s

Kidding aside, I agree, and I'm glad to see the progress that AMD has made with it. Even if it's not the absolute best, I'm interested in using Ryzen in my next system as a "vote with my wallet" for more competition.

Now if only Nvidia will hurry up with the Turing (1180) generation of Geforce... (I'd like for AMD to be more competitive in the GPU race, but they're really not at the higher end.)

[Artorius]

Kidding aside, I agree, and I'm glad to see the progress that AMD has made with it. Even if it's not the absolute best, I'm interested in using Ryzen in my next system as a "vote with my wallet" for more competition.

Technically speaking they are the absolute best now if you stay within stock configurations without overclocking. Partially due to spectre/meltdown patches which affect Intel more than AMD due to architecture differences, partially due to AMD "supporting" better RAM without it counting as an "unsupported" OC and partially because AMD actually packages a decent CPU cooler so you'll often be able to hit 4.3GHz and still stay within specifications. In other words, AMD's "default" is just better than Intel's "default".

Sure, current Intel SKUs OC higher and you can put better RAM which is "unsupported" but works well with XMP. You could also run your system without the performance hitting security patches, and in this case Intel would obviously be better. And most people spending money on something like a 8700K buy it so they can run it outside of specifications, being able to OC is a feature. So it depends on the PoV really.

What I personally find really great is that they do include a decent CPU cooler, which you would otherwise have to buy separately with Intel systems making the other choice even more expensive than it already is. Their performance/price offering is just amazing.

[Remilia]

Got bored and curious, and noticed I accidentally wrote the wrong model in the previous post.
http://direct.eizo.co.jp/shop/g/gCG3145/
This thing costs $28k~ pre-tax.

[Artorius]

I seriously wonder why hasn't Samsung hired Andrei yet...
It's honestly embarrassing that a single engineer unrelated to your company does a better job at fine-tuning your product than you do.

- - - Updated - - -

Ryan told Ian to retest a lot of things, including the old Zen SKUs on newer mobos/newer Windows/newer drivers/newer security patches. They want to be able to explain why they're getting better gaming results than basically everyone else.

I had initially thought that it was due to the spectre/meltdown patches, but apparently their Intel results are only slightly worse than everyone else's, while their Ryzen 2 results are all relatively better.

I can see 2 scenarios:

1) Perhaps they got lucky with great silicon capable of staying in 4.3GHz most of the time and are using dual ranked RAM or something else that can unexpectedly increase performance while the other reviewers aren't.
2) Retesting old Ryzen will make its numbers go up as well which isn't exactly hard to believe since the previous launch was a complete mess, the entire software stack should be much more stable now.
3) Maybe Ian fucked up and used different graphics cards in his tests...

[Remilia]

I seriously wonder why hasn't Samsung hired Andrei yet...
It's honestly embarrassing that a single engineer unrelated to your company does a better job at fine-tuning your product than you do.

Yeah, it's kind of sad. Though I'd prefer is Andrei stayed independent (relatively speaking) so we get more of these. They're nice reads. It's also why I abhor their software stack.

Ryan told Ian to retest a lot of things, including the old Zen SKUs on newer mobos/newer Windows/newer drivers/newer security patches. They want to be able to explain why they're getting better gaming results than basically everyone else.

I had initially thought that it was due to the spectre/meltdown patches, but apparently their Intel results are only slightly worse than everyone else's, while their Ryzen 2 results are all relatively better.

I can see 2 scenarios:

1) Perhaps they got lucky with great silicon capable of staying in 4.3GHz most of the time and are using dual ranked RAM or something else that can unexpectedly increase performance while the other reviewers aren't.
2) Retesting old Ryzen will make its numbers go up as well which isn't exactly hard to believe since the previous launch was a complete mess, the entire software stack should be much more stable now.
3) Maybe Ian fucked up and used different graphics cards in his tests...

I didn't really look at the other sites, but did they test at same / similar RAM configurations or at stock. Cause I know Ian always tests stock conditions first and sometimes have some OC article later on. Stock Zen+ has higher memory speeds certified at 2933MHz iirc. So that may make a difference.

[Artorius]

I didn't really look at the other sites, but did they test at same / similar RAM configurations or at stock. Cause I know Ian always tests stock conditions first and sometimes have some OC article later on. Stock Zen+ has higher memory speeds certified at 2933MHz iirc. So that may make a difference.

The others are doing all kinds of wrong things as expected, but AT can't exactly explain the difference in the results so they're running some of the things again to make sure. Some reviewers leave options in Intel mobos to auto OC turned on, some use OCd RAM, some leave spectre/meltdown mitigations turned off... It's a mess.

I posted a bunch of comments from Ryan and Ian themselves in the Ryzen 2 thread in the computer forums, but as always it's filled with retarded Intel vs AMD flaming and nobody actually cares about the technology itself or discussing things technically.

[Remilia]

The others are doing all kinds of wrong things as expected, but AT can't exactly explain the difference in the results so they're running some of the things again to make sure. Some reviewers leave options in Intel mobos to auto OC turned on, some use OCd RAM, some leave spectre/meltdown mitigations turned off... It's a mess.

I posted a bunch of comments from Ryan and Ian themselves in the Ryzen 2 thread in the computer forums, but as always it's filled with retarded Intel vs AMD flaming and nobody actually cares about the technology itself or discussing things technically.

Figured it'd be a mess.

Also yeah, kind of exactly what I expected out of mmoc.

[Alindra]

Technically speaking they are the absolute best now if you stay within stock configurations without overclocking. Partially due to spectre/meltdown patches which affect Intel more than AMD due to architecture differences, partially due to AMD "supporting" better RAM without it counting as an "unsupported" OC and partially because AMD actually packages a decent CPU cooler so you'll often be able to hit 4.3GHz and still stay within specifications. In other words, AMD's "default" is just better than Intel's "default".

Additionally, all Ryzen chips can be overclocked, while only K Intel chips can be. The benefit of this can be argued a little bit, since Ryzen doesn't seem to benefit quite as much from overclocking as Intel ones - that is, they have a lower ceiling - but it's still nice not to have to pay extra for it.

What I personally find really great is that they do include a decent CPU cooler, which you would otherwise have to buy separately with Intel systems making the other choice even more expensive than it already is. Their performance/price offering is just amazing.

I know that the included coolers was something they were making an effort to point out, but I didn't realize they were actually decent, or even good. I've been so trained that stock coolers suck that I pretty much completely dismissed them.

The others are doing all kinds of wrong things as expected, but AT can't exactly explain the difference in the results so they're running some of the things again to make sure. Some reviewers leave options in Intel mobos to auto OC turned on, some use OCd RAM, some leave spectre/meltdown mitigations turned off... It's a mess.

The heck? I was initially thinking "what are the chances that 'everyone' is doing something wrong and only one place is doing it 'right,'" but those examples are pretty egregious oversights. How the heck can benchmarks from other sites be trusted when they screw up things like leaving auto-overclocking on?

*shakes head*

[Artorius]

I don't see a problem on leaving the OC feature turned on if it comes turned on by default on the BIOS and doesn't really create any thermal problems. I only see it as a problem when they don't mention that there was an overclock, and end up reporting it as "stock" performance. Which isn't really true.

There are other things like HPET too, and a bunch of other extremely random things like the "Ryzen balanced power plan" which apparently doesn't help anymore. You can never know what reviewers did when not even them know it entirely. It would normally be safe to assume Ian got it right, but since everyone else is getting worse results in a way or another (reviews are all over the place, some put Intel on top, some have them at more or less the same level, Ian has Ryzen 2 beating everything), AT is investigating to find out why.

There's some valuable input from ComputerBase, most of the others don't include enough information for us to know what they actually did...

[Remilia]

OC features in my opinion should be relegated to MoBo reviews as it differs from cooler set up and MoBos themselves and thus doesn't judge the CPU itself.

I also really like 4Gamer's review. Can ignore it's Japanese and look at the chart if you can't, though some small things (like maximum power draw vs average) might need some translation/knowledge. They're a lot more simple compared to Anandtech but they're pretty transparent in what their test set up is like. Gaming shows 8700k in the lead though they use higher frequency RAM as denoted by the test set up, at 3400MHz speed, so not stock. Still no major discrepancies that I see the other reviews have. Results might actually match up to AT with stock RAM frequencies.
http://www.4gamer.net/games/300/G030061/20180419012/
http://www.4gamer.net/games/032/G003251/20171219123/
http://www.4gamer.net/specials/4g_be...gulation.shtml
They post their benchmark regulation. Has all of the settings and set up for how they do it. Then the review itself will always post the more variable things like BIOS version, OS version, chipset, etc etc.
They're by far the most transparent in their testing methodology.

[Artorius]

All the gaming results on Anandtech were ninja updated, the 8700K is back at the top in most titles now. The rumour is that the previous results were due to buggy nvidia drivers, but they haven't said anything about it yet. This is honestly pretty weird.