Valve sucks

[starhawk]

Some of us are still enjoying this thread, thank you.

*snort of abysmal disgust* it's thirty-three pages long, thirty-two-and-a-half of which are spam and bashing... how can ANYONE enjoy this monster?!?!?! the only thing it's gonna do if somebody doesn't kill it in the next ten pages is crash the [H] servers!

 

[R1ckCa1n]

*snort of abysmal disgust* it's thirty-three pages long, thirty-two-and-a-half of which are spam and bashing... how can ANYONE enjoy this monster?!?!?! the only thing it's gonna do if somebody doesn't kill it in the next ten pages is crash the [H] servers!

And all over 2.55% of the steam users.......

 

[starhawk]

And all over 2.55% of the steam users.......

true. this thread has to be taking megabytes of space on the main server. the thing's already getting pretty slow.

i say it's time to grab that old gravity gun and levitate this thread into a lake before Kyle has to buy a new [H]ard|Drive to fit it on.

 

[R1ckCa1n]

I think the point is that if Valve spent some time they could have made a mixed mode precision path and made playable DX9 (in the 30s) for FX users if they chose to enable it, and give a speed boost for 6800 users as well. This is what most modern intensive DX9 games have; I assume HL2 does not because its ATI sponsored and ATI would not benefit from it. There is no way in hell it is a time issue as the game was delayed a year and adding partial precision (inserting a line of code/pp hint to shaders then testing them) couldn't possibly take that long even with a game that uses a lot of shaders. The longest part would be testing them all to see which require FP32 instead of FP16, but Valve could do that in under a week - heck the shaders requiring FP32 were detected on the net in 2 days using 3danalyze.

AGAIN, for 2.55% of the users? That is not good time spend since time=money. The only time a FX5900 works good is running PS1.4 / PS2.0 mix, which is exactly what Valve said it was not going to do. Who's fault is that, NV's.

 

[tranCendenZ]

AGAIN, for 2.55% of the users? That is not good time spend since time=money.

Less than 2% of the steam users are using X800 cards, are you advocating that devs ignore SM2.0B next year once ATI has a SM3.0 card out, since the X800 is the only generation card to ever benefit from SM2.0B? As creating a sm2.0b path is far more complex than adding pp hints to shaders. If your logic sticks, the X800 will be treated as a vanilla sm2.0 card next year.

 

[R1ckCa1n]

Less than 2% of the steam users are using X800 cards, are you advocating that devs ignore SM2.0B next year once ATI has a SM3.0 card out, since the X800 is the only generation card to ever benefit from SM2.0B? And creating a sm2.0b path, btw, is far more complex than adding pp hints to shaders.

Bah, you are not going to admit Valve did no wrong by delivering EXACTLY what they said they would. You are barking up the wrong tree.

 

[tranCendenZ]

Bah, you are not going to admit Valve did no wrong by delivering EXACTLY what they said they would. You are barking up the wrong tree.

Actually Valve didn't even deliver what they said they would. They said, during ATI Shader Day 2003, that they spent 3x as long creating a mixed precision path for nv3x cards in half life 2. This was shortly before the game was leaked. Valve even provided benchmarks using this mixed precision path, which were substantially higher fps than the full precision path. That path is nowhere to be found in the game, hence they did not even deliver what they complained about in 2003.

http://www.techreport.com/etc/2003q3/hl2bench/index.x?pg=2

You can see from these old benchmarks Valve released that with the "FX" mixed precision path performance is massively higher than the full precision path. While not as fast as the 9800PRO, the mixed precision path gave the FX series a 15-20fps boost and put it in the playable range (a bit faster than a 9600PRO), coming close to the speed of the DX8.1 path. But the path was not included with the final game.

 

[starhawk]

yet this isn't about even 3% of people out there... meaning that if you have 100 people playing HL2 in the same room, then less than three people in the whole fricking room have the fricking problem.
it's pretty much pointless.

i repeat what i said before: some mod plz lock this thread before it locks the server.

 

[R1ckCa1n]

Actually Valve didn't even deliver what they said they would. They said, during ATI Shader Day 2003, that they spent 3x as long creating a mixed precision path for nv3x cards in half life 2. This was shortly before the game was leaked. Valve even provided benchmarks using this mixed precision path, which were substantially higher fps than the full precision path. That path is nowhere to be found in the game, hence they did not even deliver what they complained about in 2003.

http://www.techreport.com/etc/2003q3/hl2bench/index.x?pg=2

You can see from these old benchmarks Valve released that with the "FX" mixed precision path performance is massively higher than the full precision path. While not as fast as the 9800PRO, the mixed precision path gave the FX series a 15-20fps boost and put it in the playable range (a bit faster than a 9600PRO), coming close to the speed of the DX8.1 path. But the path was not included with the final game.

You still fail to realize it was a $$ thing. Fix the 5900 short comings or release a game. We all see what won.

 

[tranCendenZ]

You still fail to realize it was a $$ thing. Fix the 5900 short comings or release a game. We all see what won.

Oh, it was definitely a $$ thing. Had more to do with helping ATI and less to do with fixing Nvidia tho Valve had a partial precision path running in sept. 2003 which they benchmarked on several timedemos across many levels. What happened to it? They didn't have time to finish up a partial precision path they had up and running over a year before the game's release? Give me a break, that makes no sense.

 

[Impulse]

*snort of abysmal disgust* it's thirty-three pages long, thirty-two-and-a-half of which are spam and bashing... how can ANYONE enjoy this monster?!?!?! the only thing it's gonna do if somebody doesn't kill it in the next ten pages is crash the [H] servers!

It's only 22 pages long here, you need to display more posts per page to grapple with flaming behemoths such as these!

 

[rcolbert]

One could simply choose to click on the "last page" link, or even better, if you aren't enjoying the thread anymore, you could also choose to ignore it. The length of the thread isn't an issue unless you feel you might have missed something somewhere in the middle and are determined to find it.

 

[starhawk]

that's like telling someone to stop breathing when they complain that the air is thin.

plz use common sense. there's maybe 10 posts in here that are useful.

would someone please knock this thread off before it kills the server? i like [H] and i don't want to see it down because of this one stinking thread.

 

[tranCendenZ]

that's like telling someone to stop breathing when they complain that the air is thin.

plz use common sense. there's maybe 10 posts in here that are useful.

would someone please knock this thread off before it kills the server? i like [H] and i don't want to see it down because of this one stinking thread.

why are you so vested in getting a thread closed where many are finding debate and discussion useful. perhaps some of the people here want to see valve do something about the lack of a partial precision path, and don't want their thread closed and forgotten about.

For all the bugs that came out with HL2, along with Valve's arrogance in blaming it on people's machines instead of their coding, on top of the forcing of steam even for retail CDs and their purposeful discarding of the partial precision path they had running over a year ago likely to please ATI, I really have no problem with a thread entitled "Valve Sucks" being at the top of the videocard forum. Maybe one of their employees will read it and they will change their attitude and actually implement some of the things users have been asking for.

 

[Rof_lmao]

This thread sucks

 

[R1ckCa1n]

This thread sucks

And so do people who whine about 2.55% of the steam users made a bad purchase.

 

[R1ckCa1n]

Less than 2% of the steam users are using X800 cards, are you advocating that devs ignore SM2.0B

Considering the X800 is only one of many TRUE DX9 parts from ATI, this is irelivant.

next year once ATI has a SM3.0 card out, since the X800 is the only generation card to ever benefit from SM2.0B?

Yeah,,,, look what SM3.0 does for 6800 users today! w00t Oh yeah, and how HDR requires you lower your resolution just to play....

As creating a sm2.0b path is far more complex than adding pp hints to shaders. If your logic sticks, the X800 will be treated as a vanilla sm2.0 card next year.

Proof? Link?

 

[Skystalker]

AGAIN, for 2.55% of the users? That is not good time spend since time=money. The only time a FX5900 works good is running PS1.4 / PS2.0 mix, which is exactly what Valve said it was not going to do. Who's fault is that, NV's.

Okay, I simply CAN'T see one more person spout this total shit. Where did you get this 2.55 figure anyways? Pull it outa your ASS? The FX series does have a 16/32 limitation... and that is exactly 18.9% of Valve's user base!

NVidia GeForce FX 5200 Series 105,416 7.74 %
NVidia GeForce FX 5600 Series 50,560 3.71 %
NVidia GeForce FX 5700 Series 27,792 2.04 %
NVidia GeForce FX 5900 Series 26,536 1.95 %
NVidia GeForce FX 5700LE 13,491 0.99 %
NVidia GeForce FX 5900 12,289 0.90 %
NVidia GeForce FX 5500 7,267 0.53 %
NVidia GeForce FX 5950 Series 6,984 0.51 %
NVidia GeForce FX Go5200 3,461 0.25 %
NVidia GeForce FX Go5600 2,269 0.17 %
NVidia GeForce FX 5800 Series 1,482 0.11 %

18.9% NVidia FX Series
257,547 Total Cards

If each of them bought the bronze package at $49.95, the total would be $12,864,472.65

http://steampowered.com/status/survey.html

Steam cares, or should... 13 million is a lot of money. They should support the FX users if it is possible. Simply allow the option. What is the harm in that?!

God DAMN I hate it when people just start spouting off about things when they did NO research and are just out to crap on threads and trash the FX cards. STFU.

-Skystalker

 

[tranCendenZ]

Considering the X800 is only one of many TRUE DX9 parts from ATI, this is irelivant.

eh? "true" dx9? everything nv has released including the FX series is in spec and "true" dx9 too. the relevance is not whether its in spec, its that its optimal shader profile is used by the x800 series exclusively in terms of giving maximum shader performance.

Yeah,,,, look what SM3.0 does for 6800 users today! w00t Oh yeah, and how HDR requires you lower your resolution just to play....

And? SM3.0's potential hasn't even been tapped yet. That will happen next year with games like Unreal 3.

Proof? Link?

Common sense. SM2.0B doesn't support all the features of SM3.0, but it supports more than SM2.0. It doesn't have the precision of SM3.0 either. Therefore, to make a future SM3.0 game SM2.0B compliant, devs will have to do just as must testing to make sure that the lower FP24 precision of SM2.0B does not cause problems with longer SM3.0 shaders. This is identical to the testing that would have to be done with a partial precision path. Hence, the testing aspect is the same. As for the actual coding aspect, partial precision involves adding one line of code to a shader. In order to take full advantage of SM2.0B you might have to rewrite shaders to compensate for things in SM3.0 that SM2.0b does not support or will have IQ/compatibility issues with. Therefore for future games that are designed for SM3.0, SM2.0B very much will be discarded if developers use your logic, it will become the partial precision of future SM3.0 games & the x800 will go untapped like the GeForce FX due to additional coding & testing that will have to be done for the X800 that will only benefit one line of cards ever. Much cheaper and easier just to code in SM2.0 and SM3.0, and treat the X800 as SM2.0 - if 1.5% of the gaming community is using the x800, it's not worth it, right? Just like the 2.5% GeForce 5900 owners can run DX8.1 and it looks good enough, the 1.5% X800 owners can run SM2.0 and it will look good enough next year. In both cases the full potential of the cards are not tapped, but apparently to you that won't matter, as both cards need extra tweaking for their particular series that no other card needs for optimal performance.

 

[Impulse]

I think the argument on which many have based the 2.5%~ figure is the fact that a partial precision path wouldn't run very well at all on anything under the 5700 series (or maybe even that? /shrug), it still does look like 5900+ make up a bit more than 2.5% though. I dunno, with the amount of screwed up drivers and the amount of people that are probably running retail or OEM drivers instead of NVidia's it's really hard to be sure of anything.

 

[SnakEyez187]

you might have to rewrite shaders to compensate for things in SM3.0 that SM2.0b does not support or will have IQ/compatibility issues with.

And then you would have to conpensate for cards that just run PS2.0 too, so then it's what 50 percent of the market using dx9 and maybe 2 percent for those with SM3.0? I wonder which option they might care more about... And even so the majority of shaders in a game won't be SM3.0 for a long long time, so it's not such an urgent or important deal

 

[tranCendenZ]

And then you would have to conpensate for cards that just run PS2.0 too, so then it's what 50 percent of the market using dx9 and maybe 2 percent for those with SM3.0? I wonder which option they might care more about... And even so the majority of shaders in a game won't be SM3.0 for a long long time, so it's not such an urgent or important deal

The difference is that SM3.0 will be the standard until at least 2006, probably another 2-3 card generations. SM2.0B is a one card, one manufacturer generation shader model. If you were to pick three shader paths, SM3.0, SM2.0, and SM1.x would be the obvious choices. SM2.0A and SM2.0B are both "black sheep," as they are both primarily specific to single card series. Therefore next year a dev who was just coding based on percentages as Rick suggested would never code for SM2.0A/SM2.0B, because the percentage for those shader models is small and will never grow. SM1.x covers a lot of cards, SM2.0 covers a lot of card series including ones that use SM2.0A and SM2.0B, and SM3.0 will cover a lot of cards through the future. SM2.0A/SM2.0B don't, and therefore will fall into the small percentage category. So if you want to use the percentages/money argument, realize you are dooming the X800 to SM2.0 vanilla shader performance, because supporting SM2.0B just for the X800 is not much different than supporting partial precision just for the GeForce FX, if anything its more complex and more costly. A few months after ATI SM3.0 cards hit the streets next year and penetrate the market, X800 owners will be very much in the same boat as GeForce FX owners are now in terms of getting devs to maximize performance for their card, as they both need exclusive coding to get the most performance.

 

[Majeztik12]

Valve is Valve, id is id, CRYTEK is CRYTEK, ATI is ATI, nVidia is nVidia, Intel is Intel, and AMD is AMD!!!! Oh, almost forgot Microsoft is Microsoft! This code path thingy and wut not is always gonna be there. Optimizing for this and for that, will always be here. Hey were is the Glide 3D code path....uhhh never mind my ramblings. I'll just shut up while I'm ahead....GAME ON!!!!!

 

[rcolbert]

Yeah, I guess you have to be DX 9.0c compatible to be a true DX9 card. ATI should have anticipated the yet unpublished specification when they released the 9700 Pro.

 

[starhawk]

Lawd gimme strength...

 

[Optimus]

yet this isn't about even 3% of people out there... meaning that if you have 100 people playing HL2 in the same room, then less than three people in the whole fricking room have the fricking problem.

Actually, I have the problem and I'm not of that fabled 2.55%. My card pulls an average 30fps (25fps minimum) on the dx9 full precision path. This is just barely below HardOCP's definition of playable. 15-20 fps would push my framrate well into the playable range. Doom 3 runs the same shader path for all dx9 video cards and, therefore, must have a partial precision hint setup in order to get a better framerate on a FX5900 than on an ATI 9800, so what's so horrible about Valve adding such hints to the current path, considering that there's no artifacts in Doom 3 on an ATI card?

Honestly, I'm more worried about Valve's profitability than whether or not my video card runs their game oh so beautifully. If the Doom Engine can run all that Source can and can provide for a larger user base at full capabilities, then it's a no brainer for a game company to choose Doom over Source. Heck, Doom3 will run using the full capabilities of a Voodoo2 for crying outloud!... not that this exclamation really applies...

 

[Optimus]

Woohoo! The 666th post in this thread mentioned Doom 3 well after the thread had moved on from it!

 

[gordon151]

The difference is that SM3.0 will be the standard until at least 2006, probably another 2-3 card generations. SM2.0B is a one card, one manufacturer generation shader model. If you were to pick three shader paths, SM3.0, SM2.0, and SM1.x would be the obvious choices. SM2.0A and SM2.0B are both "black sheep," as they are both primarily specific to single card series. Therefore next year a dev who was just coding based on percentages as Rick suggested would never code for SM2.0A/SM2.0B, because the percentage for those shader models is small and will never grow. SM1.x covers a lot of cards, SM2.0 covers a lot of card series including ones that use SM2.0A and SM2.0B, and SM3.0 will cover a lot of cards through the future. SM2.0A/SM2.0B don't, and therefore will fall into the small percentage category. So if you want to use the percentages/money argument, realize you are dooming the X800 to SM2.0 vanilla shader performance, because supporting SM2.0B just for the X800 is not much different than supporting partial precision just for the GeForce FX, if anything its more complex and more costly. A few months after ATI SM3.0 cards hit the streets next year and penetrate the market, X800 owners will be very much in the same boat as GeForce FX owners are now in terms of getting devs to maximize performance for their card, as they both need exclusive coding to get the most performance.

The NV40, X800, NV3x, and future graphics cores support all SM2 shader model revisions. SM2B isn't a "one card, one manufacturer" thing. So, if developers were truly going to code on percentages, out of the two which do you think they would conclude has the wider support base? Also with many of the current game engines being optimized around sm2 limitations, I somehow doubt x800 users will have anything to worry about for a good while to come.

 

[Skystalker]

I think the argument on which many have based the 2.5%~ figure is the fact that a partial precision path wouldn't run very well at all on anything under the 5700 series (or maybe even that? /shrug), it still does look like 5900+ make up a bit more than 2.5% though. I dunno, with the amount of screwed up drivers and the amount of people that are probably running retail or OEM drivers instead of NVidia's it's really hard to be sure of anything.

I don't dispute any of that... Just don't spread shit where it doesn't exist. The FX series is extremely popular, and should be supported. If it can be done, reasonably, then do it. It seems to me, it could be. Let people decide if you want to sacrifice framerates for quality or not. To choose a DX9 with 16/32 should be a selectable option for all cards. If they don't want to support it, then don't. At least they have given choices.

-Skystalker

 

[Scali]

Actually, I have the problem and I'm not of that fabled 2.55%. My card pulls an average 30fps (25fps minimum) on the dx9 full precision path. This is just barely below HardOCP's definition of playable. 15-20 fps would push my framrate well into the playable range. Doom 3 runs the same shader path for all dx9 video cards and, therefore, must have a partial precision hint setup in order to get a better framerate on a FX5900 than on an ATI 9800, so what's so horrible about Valve adding such hints to the current path, considering that there's no artifacts in Doom 3 on an ATI card?

Actually, Carmack himself has mentioned in various interviews that NVIDIA does shader replacement in their drivers, for the FX series. So effectively Doom3 is not running the same shaders on all cards (if it did, it would run lots slower on FX, as Carmack had mentioned earlier in development). I'm no expert on the ARB shaders, but iirc, they have no partial precision hint support at all. Carmack originally wrote a special shader path using NVIDIA's shader extensions rather than the ARB ones, in order to improve FX performance.

Also, from what I understood, the artifacts in HL2's DX9.0 path come from the fact that it uses user clipplanes. This is something that NV hardware didn't support until the 6800 if I'm not mistaken (although even the Radeon 8500 already supported them).

Other than that I would like to remark that DX9.0c defines 32 bit precision as the minimum requirement ONLY for SM3.0. SM2.0 is still 24 bit precision, and therefore all R3x0 are still fully compliant to the standard.

Also, I would beg to differ on whether adding pp hints is less work than using SM2.0b.
These days, shaders are mostly coded in HLSL, and by simply compiling them at runtime, for the highest supported shader standard, you automatically get optimized SM2.0b shaders on X800s.

However, for pp hints, you need to rewrite the sourcecode. You could ofcourse just search-and-replace, and force the whole thing to partial precision... but if you want to do it properly, you should look at the code of each shader, and determine where you can use partial precision, and where you cannot. By forcing the entire shader to partial precision, you probably get a lot less usable shaders than by analyzing the actual sourcecode and reducing precision where possible. And doing that properly, takes a lot of time. In any case, you still have to re-test all shaders to see if they still look good enough or not (in ALL cases, not just loading one or two levels and taking a screenshot).

 

[starhawk]

Woohoo! The 666th post in this thread mentioned Doom 3 well after the thread had moved on from it!

how appropriate... both seem to involve copious amounts of h*ll on earth.

 

[Badger_sly]

Do you hear that?

Hear what?

Those are the silent black helicopters coming to get us?

Oh goody! Hopefully one of those helicopters will deliver you a copy of ATI's stock report to investors, which shows the $6 million payout to Valve.

Or you can just Google to find it.

 

[Netrat33]

So if you want to use the percentages/money argument, realize you are dooming the X800 to SM2.0 vanilla shader performance, because supporting SM2.0B just for the X800 is not much different than supporting partial precision just for the GeForce FX, if anything its more complex and more costly. A few months after ATI SM3.0 cards hit the streets next year and penetrate the market, X800 owners will be very much in the same boat as GeForce FX owners are now in terms of getting devs to maximize performance for their card, as they both need exclusive coding to get the most performance.

Before I reply: Man I can't believe this thread is still going. I'm not reading this whole thing. Just the last page

Ok. Well you're complaining about Valve not supporting GFFX users in what they can handle. If we are to believe your argument (which is mostly opinion you have no facts to back it up) We better see the same outcry/support than for users of the x800 series who can benefit from adding the extra coding.

 

[Netrat33]

The NV40, X800, NV3x, and future graphics cores support all SM2 shader model revisions. SM2B isn't a "one card, one manufacturer" thing. So, if developers were truly going to code on percentages, out of the two which do you think they would conclude has the wider support base? Also with many of the current game engines being optimized around sm2 limitations, I somehow doubt x800 users will have anything to worry about for a good while to come.

Plus we've already seen how easy it is to implement both at the same time (Farcry)
Writing code for SM3 doesn't mean the make it exclusively SM3. It still has SM2 cards to consider. So while downgrading, TranC likes to make it seem they can't stop and SM2B. Whatever.

 

[LyCoS]

If we are to believe your argument (which is mostly opinion you have no facts to back it up) We better see the same outcry/support than for users of the x800 series who can benefit from adding the extra coding.

that IS a point : if GFFX cards will benefit from PP, so can radeons.

seems like some people here bought nvda (NASDAQ) stock

 

[Kevin Lowe]

that IS a point : if GFFX cards will benefit from PP, so can radeons.

Only if the Radeon supports PP, which it won't until R5xx.

 

[Jbirney]

When ATI comes out with a Shader Model 3.0 card next year, would you advocate developers ditching SM2.0B paths & just supporting SM1.x, SM2.0, and SM3.0 due to the very small percentage of X800 generation owners, the only card series that will ever benefit from SM2.0B? After all, the coding and testing of an SM2.0B path is quite a lot of extra work for a small return (much more work than implementing partial precision, for instance), and just might not be worth it.

Scali allready dubunked your claims. Personally I don't care one way or another so long as its equal treatment to all IHV/Vendors.

Oh, it was definitely a $$ thing. Had more to do with helping ATI and less to do with fixing Nvidia tho Valve had a partial precision path running in sept. 2003 which they benchmarked on several timedemos across many levels. What happened to it?

Simalur thing happend with the NV30 path and Doom3. FX card probably can run Doom3 faster if JC had not dropped it. But I don't see you complaining about this now

I don't dispute any of that... Just don't spread shit where it doesn't exist. The FX series is extremely popular, and should be supported. If it can be done, reasonably, then do it. It seems to me, it could be. Let people decide if you want to sacrifice framerates for quality or not. To choose a DX9 with 16/32 should be a selectable option for all cards. If they don't want to support it, then don't. At least they have given choices.

-Skystalker

Again people have said and the data shows that the 5200/5600 cards just can not run it in DX9 even with using the 16pp hack. The user said he on his 5200 he still had LOW DOUBLE DIGIT FRAME RATES. The for all of these users (which make up the majority of the total FX Steam numbers) using PP still wont help them. Why should Valve spend time on the small precentage thats left? If you want them to spend time there, what about adding PS3.0 to help all the 6800 users? What about adding 3Dc for all the 8xx ATI users? HDR? What your failing to understand is they could not add everything. The looked at their numbers and made a call. So far it looks like it was justifible given the data we have..

 

[pxc]

Again people have said and the data shows that the 5200/5600 cards just can not run it in DX9 even with using the 16pp hack. The user said he on his 5200 he still had LOW DOUBLE DIGIT FRAME RATES.

I recapped just for you and you still didn't understand what I wrote.

I
used
a
PCI
5200.

AGP
is
faster.

hack
with
ATI
"DX8.1"
path
enables
world
reflections
and
is
playable
on
even
a
PCI
5200.

is that slow enough for you now?

 

[Jbirney]

I recapped just for you and you still didn't understand what I wrote. :rolleyes

There is no reason to be an butt head about this. I understood what you wrote the first 3 times you wrote it. Yes AGP is faster.

But BY YOUR OWN words you stated that its not playable using the PP hack. You even further stated that its not going to give you playable framerates unless you use a 5700+ using the PP hack. This the majority of the FX users in steam still wont have playable DX9 using this PP hack. Thats the whole point of this useless thread right? Sure you found other ways to make it playable and thats good. Now let it go.

 

[Optimus]

I think pxc is saying that it's annoying that you get better performance regardless of shader path if you fool HL2 into believing that you have an ATI card. I'm still a little skeptical on whether you're actually making HL2 treat it like an ATI card, though. Granted, I have not yet tested it myself.