X1950xtx

sephiroth said:
I thought the gains were going to be less than that because of kentfield's slow CMD bus and data FSB.. but I guess not. If kentsfield can get away with those numbers with it's cramped CMD bus then I can't even imagine what K8L is going to turn out.
Please note that the Kentsfield FSB will probally be 1333MHz not 1066MHz.
And of course they overclock like mad.
The 4X4 is an unknown quantity we will have to wait and see.
I wish AMD luck and it is kind of ironic that Dell finally goes with them after they have just recently lost the performance edge they have enjoyed over Intel for the last few years.

BTW I have nothing against Nvidia. I have owned and will own every topend/middle of the road card from every manufacturer that I can. Anyone rember the Kyro 2?
STMicro has now been tasked by Intel to make a "real" 3d core for their upcoming chipsets. They are a kickass little British company and can be with Intel's deep pockets a force to be reckoned with.
They made a Kyro 3 that would have been nice bang for buck wise at the time but it never came out.
 
Last edited:
yashart@work said:
;) I'm at work, so im scared to post much of anything. There's alot I would love to tell you guys, especially about AMD purchasing us.... but I simply can't.

As for reference boards, we have several laying around, but not too much I can tell ya about them :(
Ah you dong...Just go ahead an tease us but I have my sources so I bet I could give it a good guess :)

I do know your next core will have 64 PIXEL SHADERS/PIPES (omg)(shocked)(omg)
Sweet mother of Moneky nads!!!!
I will have to go look on my NDA cheatsheet plus it may gobble down 200WATTS...OMG can you say external power source?
 
Last edited:
yashooa said:
Please note that the Kentsfield FSB will probally be 1333MHz not 1066MHz.
And of course they overclock like mad.
The 4X4 is an unknown quantity we will have to wait and see.
I wish AMD luck and it is kind of ironic that Dell finally goes with them after they have just recently lost the performance edge they have enjoyed over Intel for the last few years.
the data bus which you describe is not what I'm concerned about. Kentsfield is using a FSB arch that dates back to the Pentium pro.. it consists of two pathways, a quad-pumped data path (that runs @ 266mhz in the case of 1.06ghz FSB cpus, and 333mhz in the case of 1.3ghz FSB cpus) , but the CMD pathway has been left alone.. it still is running at a Single data rate of 266 or 333mhz... If Kentsfield were using HTT vs the P5 FSB arch you'd see an instant 20-80% across the board gain in every application I'd bet.

I do know your next core will have 64 PIXEL SHADERS/PIPES
G80 uses 128 unified shadders ;)
RD600 dosen't really have 64 pixel shadders, since it uses a USA (unified shadder arch) it has dynamic "pipelines" that can adjust to become vertex or pixel shadders, depening on the application. The same goes with the X1900(it uses 48 USA shadders), and G80
 
Last edited:
Geeks unite!...sorry...just had to say it as I too am one of them. Nice systems all.

Side note, should this not be in off topic?
 
sephiroth said:
the data bus which you describe is not what I'm concerned about. Kentsfield is using a FSB arch that dates back to the Pentium pro.. it consists of two pathways, a quad-pumped data path (that runs @ 266mhz in the case of 1.06ghz FSB cpus, and 333mhz in the case of 1.3ghz FSB cpus) , but the CMD pathway has been left alone.. it still is running at a Single data rate of 266 or 333mhz... If Kentsfield were using HTT vs the P5 FSB arch you'd see an instant 20-80% across the board gain in every application I'd bet.


G80 uses 128 unified shadders ;)
RD600 dosen't really have 64 pixel shadders, since it uses a USA (unified shadder arch) it has dynamic "pipelines" that can adjust to become vertex or pixel shadders, depening on the application. The same goes with the X1900(it uses 48 USA shadders), and G80
Holy hell, good post. Yeah I know goes back to the Pentium Pro 150 way way way back -->P2-->P3-->Tulatin-->then off to the Jews to become the Banias-->then Intel says, ah netburst is to long in the pipeline and too hot let's kill it.-->and the old school becomes new again.
I know about the unified shader stuff to.
Good thing AMD stold some Dec-Alpha guys to help with their first Athlon EV6 bus.

<table class="wikitable"> <tbody><tr><th>Processor Class</th> <th>FSB Frequency</th> <th>FSB Type</th> <th>Theoretical Bandwidth</th> </tr> <tr> <td>Pentium II</td> <td>66/100 MHz</td> <td>GTL+</td> <td>533/800 MB/s</td> </tr> <tr> <td>Pentium III</td> <td>100/133 MHz</td> <td>GTL+</td> <td>800/1066 MB/s</td> </tr> <tr> <td>Pentium 4<small>*</small></td> <td>100/133/200 MHz</td> <td>AGTL+</td> <td>3200/4266/6400 MB/s</td> </tr> <tr> <td>Pentium M<small>*</small></td> <td>100/133 MHz</td> <td>AGTL+</td> <td>3200/4266 MB/s</td> </tr> <tr> <td>Pentium D<small>*</small></td> <td>133/200 MHz</td> <td>AGTL+</td> <td>4266/6400 MB/s</td> </tr> <tr> <td>Pentium 4 EE<small>*</small></td> <td>200/266 MHz</td> <td>AGTL+</td> <td>6400/8533 MB/s</td> </tr> <tr> <td>Intel Core<small>*</small></td> <td>133/166 MHz</td> <td>AGTL+</td> <td>4266/5333 MB/s</td> </tr> <tr> <td>Intel Core 2<small>*</small></td> <td>166/200/266 MHz</td> <td>AGTL+</td> <td>5333/6400/8533 MB/s</td> </tr> <tr> <td>Xeon - P6 core</td> <td>100/133 MHz</td> <td>GTL+</td> <td>800/1066 MB/s</td> </tr> <tr> <td>Xeon<small>*</small> - Netburst core</td> <td>100/133/166/200/266 MHz</td> <td>AGTL+</td> <td>3200/4266/5333/6400/8533 MB/s</td> </tr> <tr> <td>Xeon<small>*</small> - Woodcrest core</td> <td>266/333 MHz</td> <td>AGTL+(with Dual Independent Buses)</td> <td>17066/21333 MB/s</td> </tr> <tr> <td>Athlon/Duron<small>**</small></td> <td>100/133 MHz</td> <td>EV6</td> <td>1600/2133 MB/s</td> </tr> <tr> <td>Athlon XP/Sempron<small>**</small></td> <td>133/166/200 MHz</td> <td>EV6</td> <td>2133/2666/3200 MB/s</td> </tr> <tr> <td>Athlon 64/X2/Opteron<small>***</small></td> <td>600/800/1000 MHz</td> <td>Hypertransport</td> <td>7500/12800/14400 MB/s</td> </tr> <tr> <td>PowerPC 970****</td> <td>450/500/625 MHz</td> <td>Elastic</td> <td>7200/8000/10000 MB/s</td> </tr> <tr> <td colspan="4" rowspan="4">Notes:
<small>* - Pentium 4, Pentium M, Pentium D, Pentium EE, Xeon, Intel Core, and Intel Core 2 processors use a front side bus that transfers data four times per cycle


(omg)</small>NNvidia's upcoming G80 chip should include many things such as 1.35GHz scalable clock for 128 shader units, and a 384-bit memory controller.</td></tr></tbody> </table>
 
Last edited:
Xeon* - Woodcrest core 266/333 MHz AGTL+(with Dual Independent Buses) 17066/21333 MB/s
thank god for this one.. I would have had to slap Intel if they didn't give each Xeon cpu it's own FSB to the NB this time around..
 
HAHAHAHAH
BTX is dead...
THE WORD IS OUT among people in the channel. According to next year's line-up, the Intel has decided not to release a single BTX motherboard. Current BTX and picoBTX motherboards will be replaced with ATX and ATX form factors.
 
atticus1398 said:
wasnt BTX just a new form factor so Intels PresHOT wouldnt melt?
Not sure but I know that Intel was hoping for broader adoption of the standard. Just thought it was funny no one else wanted it.
 
Last edited:
Yes, I am 99% sure that Intel invented that form factor so that the Presscott chip wouldnt burn down your house.
 
atticus1398 said:
Yes, I am 99% sure that Intel invented that form factor so that the Presscott chip wouldnt burn down your house.
They way that it was pitched to us was that it was a brave new layout that was far superior to ATX. But the Pressy was and is indeed a hot b****. I could not believe that those things were still working at 80c!


Check this out.

"WE LEARNED that the upcoming G80 can render a real chick. It is so powerful that it can put Nalu and Dawn, the previous Geforce, models to shame. Nvidia hired a professional model, a cool chip chick, who featured in the February 2006 issue of Playboy, and did a live session with her. They captured her motion and shot supporting photography.
The demo for the big launch in 'week 45' or early November will show a super-hot chick based on a genuine celebrity model, not a made-up figure, and it will probably make you drool.
Clips of the actual chick have been mixed by Nvidia's demo boys with some motion capture from a generic ballet dancer.
All the sessions were performed recently in you know where, southern California. With the power of the G80, Geforce 8800 GTX, it has to look great."
 
yashooa said:
HAHAHAHAH
BTX is dead...
THE WORD IS OUT among people in the channel. According to next year's line-up, the Intel has decided not to release a single BTX motherboard. Current BTX and picoBTX motherboards will be replaced with ATX and ATX form factors.
omfg are you kidding?! (shocked)


it would make sense, since BTX was supposed to be the platform for tejas(the 10ghz, 200W prescott), and since conroe is yards better on thermal management than even prescott was..


eitherway, ATX is (first)
 
sephiroth said:
omfg are you kidding?! (shocked)


it would make sense, since BTX was supposed to be the platform for tejas(the 10ghz, 200W prescott), and since conroe is yards better on thermal management than even prescott was..


eitherway, ATX is (first)

Man that 200W chip would have been just crazy.
 
Forgot about this one X1800XL AIW.
Was in the closet.
trans%2010659s.jpg
 
if your are'nt coming up off of any of these, then stop posting pics and making me jealous!lol
 
whats the deal? I know you get "reference samples" but why let them pile up dust? Do they eventually ask for them back? Or are you going to one day make the worlds first 30 card crossfire system?
 
oh yeah, will ATI be venturing into the Physics Card realm? Feel free to pm any juicy AMD etc. info to me.:)
 
Back