GeneticWeapon
Splendid
- Jan 13, 2003
- 5,795
- 0
- 25,780
He's right Dave, it doesnt work...
3DMark 03 = 4,140
<A HREF="http://service.futuremark.com/compare?2k3=897633" target="_new">http://service.futuremark.com/compare?2k3=897633</A>
<font color=red>AthlonXP 2100+/Radeon 9500Pro</font color=red>
<font color=red>Folding for Beyond 3D</font color=red>
- Feb 18, 2003
- 20,043
- 0
- 50,960
To me, multi-card-multi-monitor support. Just think 3 card running 3 monitors at 24x20 (or something) that's what I want.
I agree though the bus speed itself wont be utilised for a while as the 5900U and 9800-256 barely see any diff. from what I've read. But it's the future that's the issue. Would you have felt more secure if instead of 8X it was 16X that came out? Yeah we still don't use the full potential, but less worry that that potential might run out. I would have prefer 8X AND 16X support in motherboards, but the release of only 8X cards. However I think the future was pretty much written a while back with 3GIO and thus AGP didn't need to look too far forward.
- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <font color=green>RED</font color=green> <font color=red>GREEN</font color=red> :tongue: GA to SK
I agree though the bus speed itself wont be utilised for a while as the 5900U and 9800-256 barely see any diff. from what I've read. But it's the future that's the issue. Would you have felt more secure if instead of 8X it was 16X that came out? Yeah we still don't use the full potential, but less worry that that potential might run out. I would have prefer 8X AND 16X support in motherboards, but the release of only 8X cards. However I think the future was pretty much written a while back with 3GIO and thus AGP didn't need to look too far forward.
- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <font color=green>RED</font color=green> <font color=red>GREEN</font color=red> :tongue: GA to SK
- Feb 18, 2003
- 20,043
- 0
- 50,960
If you read the article I posted and the title of the thread. IT mentions there will be an AGP to PCI-EX adapter shipping with the cards, which is necessary to allow them to work, since there appears to be no Mobo's availible at/around the target release date.
That might come in handy in the future too. I wonder if they will be common place or on of those things that someone has 'in the back' to test 'old' mother boards with the 'current' cards.
It should be backwards compatible, but the question is with what? 8X only? 2/4/8X? Dunno!
If you meant the other way around, (running AGP in PCI-EX) not without an adapter, and I doubt they would make the effort to do that as the market would be VERY limited and with no gain. But you definitely can't just plug it in.
- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <font color=green>RED</font color=green> <font color=red>GREEN</font color=red> :tongue: GA to SK
That might come in handy in the future too. I wonder if they will be common place or on of those things that someone has 'in the back' to test 'old' mother boards with the 'current' cards.
It should be backwards compatible, but the question is with what? 8X only? 2/4/8X? Dunno!
If you meant the other way around, (running AGP in PCI-EX) not without an adapter, and I doubt they would make the effort to do that as the market would be VERY limited and with no gain. But you definitely can't just plug it in.
- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <font color=green>RED</font color=green> <font color=red>GREEN</font color=red> :tongue: GA to SK
davepermen
Distinguished
- Sep 7, 2002
- 386
- 0
- 18,780
okay, fixed the link
yes, there is much gain: agp is slow at streaming FROM gpu TO mainmemory/cpu
graphic-cards today are oneway streets. you do first part on cpu, second on gpu. but then you can not do anything anymore with the result on the cpu. this is often desirable, and technically no problem. problem is that readbacks from the gpu are rather slow. while you can stream millions of triangles over to the gpu, drawing, you cannot stream back for example the full image, process it, and then for example store it in a movie-file. even while all the other hw would be about fast enough..
or, said in another way: who actually needs gpu-ram? i mean.. the gpu ram is just a buffer because streaming over agp is too slow. cpu-ram would be a much bigger and like that bether place to store the geometry. as the cpu needs the geometry anyways, for physics, for precalculations, for animating (parts of the logic of animation will always happen on cpu's), etc. so most games store meshes twice.
why? because agp sucks.
this gets espencially true for raytracing. and it is comming. more and more. i can't wait..
agp is a bottleneck, even today. it restricts us from new renderingmethods. its a design bottleneck.
pci-express will just drop away _ANY_ possible bus-bottlenecks existing. thats always cool.
now, after reading, you can check out my pics
<A HREF="http://devmaster.net/forums/index.php?act=ST&f=29&t=268&s=6405213afacf89cfa97db8fd202b7114" target="_new">page on devmaster</A>
"take a look around" - limp bizkit
www.google.com
yes, there is much gain: agp is slow at streaming FROM gpu TO mainmemory/cpu
graphic-cards today are oneway streets. you do first part on cpu, second on gpu. but then you can not do anything anymore with the result on the cpu. this is often desirable, and technically no problem. problem is that readbacks from the gpu are rather slow. while you can stream millions of triangles over to the gpu, drawing, you cannot stream back for example the full image, process it, and then for example store it in a movie-file. even while all the other hw would be about fast enough..
or, said in another way: who actually needs gpu-ram? i mean.. the gpu ram is just a buffer because streaming over agp is too slow. cpu-ram would be a much bigger and like that bether place to store the geometry. as the cpu needs the geometry anyways, for physics, for precalculations, for animating (parts of the logic of animation will always happen on cpu's), etc. so most games store meshes twice.
why? because agp sucks.
this gets espencially true for raytracing. and it is comming. more and more. i can't wait..
agp is a bottleneck, even today. it restricts us from new renderingmethods. its a design bottleneck.
pci-express will just drop away _ANY_ possible bus-bottlenecks existing. thats always cool.
now, after reading, you can check out my pics
<A HREF="http://devmaster.net/forums/index.php?act=ST&f=29&t=268&s=6405213afacf89cfa97db8fd202b7114" target="_new">page on devmaster</A>"take a look around" - limp bizkit
www.google.com
GeneticWeapon
Splendid
- Jan 13, 2003
- 5,795
- 0
- 25,780
Fuuck Dave, I thought you were much, much,much older.
Fuuck your smart for your age.
I must have grew up in the wrong damn country....
Very nice of you to share the pictures with us. It feels like we know you a little better now. Before, you were a little bit of an "unknown"
3DMark 03 = 4,140
<A HREF="http://service.futuremark.com/compare?2k3=897633" target="_new">http://service.futuremark.com/compare?2k3=897633</A>
<font color=red>AthlonXP 2100+/Radeon 9500Pro</font color=red>
<font color=red>Folding for Beyond 3D</font color=red>
Fuuck your smart for your age.
I must have grew up in the wrong damn country....
Very nice of you to share the pictures with us. It feels like we know you a little better now. Before, you were a little bit of an "unknown"
3DMark 03 = 4,140
<A HREF="http://service.futuremark.com/compare?2k3=897633" target="_new">http://service.futuremark.com/compare?2k3=897633</A>
<font color=red>AthlonXP 2100+/Radeon 9500Pro</font color=red>
<font color=red>Folding for Beyond 3D</font color=red>
davepermen
Distinguished
- Sep 7, 2002
- 386
- 0
- 18,780
hehe, no.. not yet 45.. (and happy about that)
in fact, i just bought my first car (nearly as old as i am, he's 14, i'm 19), still in education (learning IT specialist, working most of the week, 2 days per week school.. aprenticeship or however that thing's called), just about 1.5 months now in love with my GF, etc.
not sure about the future yet. but as lont as i cannot render GI realtime (global illumination), i'm not happy yet. well, i _am_ happy, very happy. but not with renderingquality. technically, _no_ current engine could dynamically light ceilings if all they could have would be lights shining from ceilings downwards (as, interestingly, about all indoor scenes do have, namely buildings..). they would all be black ceilings.
does that hint we are currently rendering with sucking technology? does it?
if so, support raytracing and gi
too bad my gf is taking so much time that i nearly don't have time for coding (well, bad for the coding anyways, for me, its great, HAHA)
how old are you, dudes?
"take a look around" - limp bizkit
www.google.com
)in fact, i just bought my first car (nearly as old as i am, he's 14, i'm 19
), still in education (learning IT specialist, working most of the week, 2 days per week school.. aprenticeship or however that thing's called), just about 1.5 months now in love with my GF, etc.not sure about the future yet. but as lont as i cannot render GI realtime (global illumination), i'm not happy yet. well, i _am_ happy, very happy. but not with renderingquality. technically, _no_ current engine could dynamically light ceilings if all they could have would be lights shining from ceilings downwards (as, interestingly, about all indoor scenes do have, namely buildings..). they would all be black ceilings.
does that hint we are currently rendering with sucking technology? does it?
if so, support raytracing and gi
too bad my gf is taking so much time that i nearly don't have time for coding (well, bad for the coding anyways, for me, its great, HAHA)
how old are you, dudes?
"take a look around" - limp bizkit
www.google.com
GeneticWeapon
Splendid
- Jan 13, 2003
- 5,795
- 0
- 25,780
I'll be 34 on the 24th of this month
3DMark 03 = 4,140
<A HREF="http://service.futuremark.com/compare?2k3=897633" target="_new">http://service.futuremark.com/compare?2k3=897633</A>
<font color=red>AthlonXP 2100+/Radeon 9500Pro</font color=red>
<font color=red>Folding for Beyond 3D</font color=red>
3DMark 03 = 4,140
<A HREF="http://service.futuremark.com/compare?2k3=897633" target="_new">http://service.futuremark.com/compare?2k3=897633</A>
<font color=red>AthlonXP 2100+/Radeon 9500Pro</font color=red>
<font color=red>Folding for Beyond 3D</font color=red>
daddywags214
Distinguished
- Mar 30, 2003
- 939
- 0
- 18,980
AGP does suck for raytracing. I've had experience... oh wait, it could just be my crappy machine. Yes, that's it.
Tit for tat, butter for fat, ATi's dog kicks nVidia's cat
(Maximum PC)
Tit for tat, butter for fat, ATi's dog kicks nVidia's cat
(Maximum PC)
davepermen
Distinguished
- Sep 7, 2002
- 386
- 0
- 18,780
interesting note: a simple triangle, raytraced, on 320x240, runs on my configuration at about 140fps.. that is
celeron 2gig (p4 based), ddr 333mhz, fsb 133mhz.
on my friends system:
3.2gig p4 ht, ddr 400mhz, fsb 800mhz: 700fps!!
thats a factor 5!!!!
next he'll run realstorm on it (www.realstorm.com), so i'll see how good the system is on that one..
this shows how a fast bus (fsb in this case) can really speed up by far the performance for even simple raytracing. agp sucks as it is very slow in one direction. can't wait for pci-express cards.. it will really help future development!!!
"take a look around" - limp bizkit
www.google.com
celeron 2gig (p4 based), ddr 333mhz, fsb 133mhz.
on my friends system:
3.2gig p4 ht, ddr 400mhz, fsb 800mhz: 700fps!!
thats a factor 5!!!!
next he'll run realstorm on it (www.realstorm.com), so i'll see how good the system is on that one..
this shows how a fast bus (fsb in this case) can really speed up by far the performance for even simple raytracing. agp sucks as it is very slow in one direction. can't wait for pci-express cards.. it will really help future development!!!
"take a look around" - limp bizkit
www.google.com
eden
Champion
- Oct 3, 2001
- 26,549
- 0
- 50,780
There are a few anomalies in your interpretation here:
-For one, there is no 3.2GHZ HT, it must be some overclocked one, because AFAIK Intel has not released any, and if he did, it must be a sample from them.
-True, the bus went from 133 to 200, but I would not attribute only that to the higher performance. Keep in mind a 2GHZ Celeron is about as good as a 1.2GHZ Thunderbird. The same clocked P4 at 512K L2, has about 30% more performance. Therefore one at 3GHZ and higher FSB, would literally rape it, not to mention HyperThreading presence, in the event of optimized threading. You also have the presence of likely a Springdale or Canterwood chipset which boosts even more the 800MT performance!
--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:
-For one, there is no 3.2GHZ HT, it must be some overclocked one, because AFAIK Intel has not released any, and if he did, it must be a sample from them.
-True, the bus went from 133 to 200, but I would not attribute only that to the higher performance. Keep in mind a 2GHZ Celeron is about as good as a 1.2GHZ Thunderbird. The same clocked P4 at 512K L2, has about 30% more performance. Therefore one at 3GHZ and higher FSB, would literally rape it, not to mention HyperThreading presence, in the event of optimized threading. You also have the presence of likely a Springdale or Canterwood chipset which boosts even more the 800MT performance!
--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:
davepermen
Distinguished
- Sep 7, 2002
- 386
- 0
- 18,780
there is a 3.2gig ht. nobody really cared yet, but at least its buyable. (and i've seen it on the right side of th noted a while ago..)
possibly this contributes to the higher performance, due the quite big amount of memory i write and read the whole time..
hyperthreading doesn't help, as it runs in one thread, my app.
"take a look around" - limp bizkit
www.google.com
possibly this contributes to the higher performance, due the quite big amount of memory i write and read the whole time..
hyperthreading doesn't help, as it runs in one thread, my app.
"take a look around" - limp bizkit
www.google.com
davepermen
Distinguished
- Sep 7, 2002
- 386
- 0
- 18,780
don't tell me i've seen it yet, but his sources are.. well.. special (possibly it _is_ a special production.. he always has the new stuff first.. and i don't even have the old stuff hehe) (btw, he has 2016 RayMarks, i've only 771 .. (www.realstorm.com))
"take a look around" - limp bizkit
www.google.com
i've seen it yet, but his sources are.. well.. special (possibly it _is_ a special production.. he always has the new stuff first.. and i don't even have the old stuff hehe) (btw, he has 2016 RayMarks, i've only 771 .. (www.realstorm.com))"take a look around" - limp bizkit
www.google.com
TRENDING THREADS
-
-
News Nvidia's grasp of desktop GPU market balloons to 88% — AMD has just 12%, Intel negligible, says JPR- Started by Admin
- Replies: 79
-
-
Question Recently Bought SSD with low health status, abnormal stats on crystaldisk?
- Started by Salads
- Replies: 9
-
Facebook
Twitter
Instagram