TechSpot adopts latency benchmarks in new Titan review

[Annisman*]

This is the most entertainment I've had all week! Just wanted to get in and be a part of the history before it gets locked.

 

[Will Robinson]

MTDEW
Diamond Member



Join Date: Oct 1999
Location: Central PA
Posts: 3,690

Quote:
Originally Posted by blackened23
I'm willing to bet that you verbally argue with traffic signs.

That there is a sig worthy statement if i ever saw one. :biggrin:

This one is just pwns LOL

Originally Posted by BFG10K
Have you considered your own blog with comments disabled?

But this is QFT:

Quote:
Originally Posted by lavaheadache
"My personality is who I am...my attitude depends on who YOU are"- Lawnberger
I don't think who we are will make one iota about your attitude

Must be something about folks from Denmark. Both him and ShintaiDK exhibit the same open hostility towards forum-goers.

 

[3DVagabond]

Pure min/max numbers are pointless.

If your chugging along at 60fps 99% of the time but hit 2fps for a short time due to a hitch and 300fps for a few frames because you just stared at the sky box would those min/max numbers be at all useful?

No point in explaining anything to him.

 

[3DVagabond]

^_^

Again, it won't make any difference. It's the same as pointing out that 7970's aren't $550 any more.

 

[3DVagabond]

So is the whole "Radeon cards stutter more" myth busted? According to the Techspot article it seems like GTX 680/HD 7970 Ghz are pretty much the same when it comes to frame latency.

I know Crossfire is still at a disadvantage compared to SLi but last month there was a lot of rambling about single Radeon cards being stuttery.

And we still have people pointing to the two different reports and combining them together to maintain that image. They think they are smarter than everyone else and nobody will notice.

I agree with those who want to see graphs. Just like posting a single number for min/max isn't useful, I don't see how posting a single 99% number is any more useful.

Edit: We also need to test at what point the latency becomes noticeable. I'm pretty sure we'll see that number moves depending on the current frame rate. There is likely some combined graph that would be meaningful as a single reported reading.

 

[BrightCandle]

Pcperspective have shown you can't benchmark sli and xfire with frame times exclusively. Anyone that does isn't paying attention.

Anandtech, sigh. 3 months is a long time to be absent in this, to have not evolved past 1 fps number, I mean wow you missed the marek entirely.

Worthless GPU reviews are worthless.

 

[3DVagabond]

Wow the attempt at intimidation is so blatant. LOL

 

[Grooveriding]

Pcperspective have shown you can't benchmark sli and xfire with frame times exclusively. Anyone that does isn't paying attention.

Anandtech, sigh. 3 months is a long time to be absent in this, to have not evolved past 1 fps number, I mean wow you missed the marek entirely.

Worthless GPU reviews are worthless.

All PCPER's results are suspect until he comes forward on who the unknown third party is that is giving him his tools and guidance for his results : Ryan has been helping a very big industry player to test a toolset that can capture every frame coming out of a graphics card over a DVI connection and then analyze frame delivery times. | His big industry partner has been saying similar things about Fraps not being "entirely accurate" to review site editors behind the scenes for some time now.

Sure I guess it could be Asus, Dell or some other monitor manufacturer. But there is a large possibility it would have to be one of the only two companies making consumer video cards. Considering the poor light his results put AMD in, I think it's safe to assume if it is a GPU maker, it's nvidia.

I can't trust anything he's saying so long as he doesn't see fit to be up front on who is providing him this tool he is using and the advice and guidance on the relevance of FRAPS etc. It would be one thing to release data with the disclaimer he's working with nvidia, it's totally another to be releasing his findings like he is without disclosing who the 'big industry player' is that is helping him.

Why is he not being transparent about this ? That is not how you conduct any sort of credible research, when your results are going to be beneficial/detrimental to a company or company's competitor that is involved in your work...

 

[Ryan Smith]

I've just got to ask, where is Anandtech on all of this?

Difficult to see. Always in motion is the future.

We have something in the works right now. Something I believe will satisfy all of you. But it isn't ready today. Soon, my friends. Soon.

 

[3DVagabond]

Difficult to see. Always in motion is the future.

We have something in the works right now. Something I believe will satisfy all of you. But it isn't ready today. Soon, my friends. Soon.

Ryan, You give us something when it's ready for prime time. We don't need any half baked process that might or might not give us good pertinent info. Right now all we have are some numbers that are obviously polarizing in their effect.

 

[Gikaseixas]

Difficult to see. Always in motion is the future.

We have something in the works right now. Something I believe will satisfy all of you. But it isn't ready today. Soon, my friends. Soon.

Thanks for the update. I'm ok with this, rather wait for a near perfect method than a half backed one. This only show's how good Anandtech.com really is. They don't copy others just because it's the new trend, they go one step further

 

[Keysplayr]

Thanks for the update. I'm ok with this, rather wait for a near perfect method than a half backed one. This only show's how good Anandtech.com really is. They don't copy others just because it's the new trend, they go one step further

It does? Because as of now, it doesn't show anything like that. After AT finally jumps in and it does prove to go a step further than TR or PCPer, then we can say this. We have no idea what AT is cooking up. I don't want to be let down again.

 

[Termie]

Difficult to see. Always in motion is the future.

We have something in the works right now. Something I believe will satisfy all of you. But it isn't ready today. Soon, my friends. Soon.

I trust this new revelation will bring balance to the Force. The young Jedi are restless.

 

[pmv]

Pure min/max numbers are pointless.

If your chugging along at 60fps 99% of the time but hit 2fps for a short time due to a hitch and 300fps for a few frames because you just stared at the sky box would those min/max numbers be at all useful?

Well I've always thought 'max' was an entirely useless figure, but if you are comparing a standard run-through of a demo or something then including minimums is better than only having the average, I would have thought. It gives _some_ idea how bad it might get in the more demanding bits of a game.

I'm sure I vaguely remember when Tom's first started including minimums and even then I thought the ideal situation would be some sort of distribution curve showing how much time was spent at each given FPS - which is pretty much what we now seem to be moving to, no? This seems to be a good thing.

 

[Imouto]

Pcperspective have shown you can't benchmark sli and xfire with frame times exclusively. Anyone that does isn't paying attention.

Anandtech, sigh. 3 months is a long time to be absent in this, to have not evolved past 1 fps number, I mean wow you missed the marek entirely.

Worthless GPU reviews are worthless.

You should stop trampling every single thread with this stuff. First TR was the best by doing latency tests. IT took you more than 1 year to notice just because they did a very harsh and sloppy review showing the 7950 under a really bad light. Once other sites started to do the same tests and proved that AMD and NV are almost the same you just went for the PCPer testing. Talk about credibility.

It's been 2 weeks already since PCPer said that they're doing the so called holy grail of frame latency testing. I bet they're having a hard time because they will debunk their own testing as they promised to do vsync enabled and frame limiting testing. That or they will show even more loopholes than in its previous testing.

Also, you're just making fun of yourself as shown in a recent thread with people laughing at you for 2 whole pages.

 

[Gikaseixas]

It does? Because as of now, it doesn't show anything like that. After AT finally jumps in and it does prove to go a step further than TR or PCPer, then we can say this. We have no idea what AT is cooking up. I don't want to be let down again.

Yes it does. Just because they didn't copy others right away (like you would like them to and we all know why) doesn't mean that it is a negative. He was very clear on why AT seems to be behind now, they're investigating a new method, something new and maybe a change the entire industry will embrace.

 

[blackened23]

On the other hand, some of the other sites that were early to adopt frame time latency testing are constantly attacked for having improper testing methods. I mentioned some Tomshardware benchmarks here, in addition to PCPer benchmarks that showed SLI having significantly better frametimes than single cards. What was the response? Criticism. "Tomshardware is a joke", "PCPerspective has no idea how to test!" These are all things i've seen posted here.

Still to this day there are tons of websites showing the GTX 690 to have better frametimes than the Titan. Yet, can these results be accurate? According to some here, of course not. Nevermind giving a proper explanation as to why, but because the results don't fit pre-conceived notions these websites are criticized. There are many websites using frame latency testing that you've criticized just as much, if not more, than AT because of this very reason. Because the results don't fit your expectations. Maybe before jumping down Ryan's throat here you should take that into consideration - you're quick to judge tomshardware and PCperspective for having what is, in your opinions, bad frametime testing YET at the same time you jump on Ryan for not adopting it. That sounds very backwards.

He said it's going to happen. Let him take the time to develop a sound and accurate frametime latency testing method - if he doesn't do this, you will just criticize him. It's a damned if you do, damned if you don't catch-22, some people will just complain either way. Titan was just released, and i'm sure for the next big review it will be there.

 

[Keysplayr]

Yes it does. Just because they didn't copy others right away (like you would like them to and we all know why) doesn't mean that it is a negative. He was very clear on why AT seems to be behind now, they're investigating a new method, something new and maybe a change the entire industry will embrace.

^ IN BOLD was an infantile statement. Why would I want anything like that?

As to your overall optimism that Anandtech will shine
above the rest when they join the frametime wagon?
I'll believe it when I see it.