Seems the X1900 XTX are cherry picked cores

[apoppin]

Originally posted by: sindows

Originally posted by: apoppin

Originally posted by: M0RPH
I suspected this might be the case. For someone that really wants to overclock the hell out of their card, the XTX might just be worth it.

help me with this a little

why does someone buy the single fastest GPU for $650 and then o/c it?

Since when did people not ever not take advantage of someting being free? I don't know about you rich people but as a student, I take advantage of all "freebies" that I can.

O/C'ing is not "free"

you 'pay' for it in loss of warranty and shortened life of your card

 

[deadseasquirrel]

Originally posted by: apoppin
edit: OMG, i just agreed with Rollo :Q

Live pictures from hell.

 

[nRollo]

Originally posted by: deadseasquirrel

Originally posted by: apoppin
edit: OMG, i just agreed with Rollo :Q

Live pictures from hell.

:laugh:

 

[NoStateofMind]

Originally posted by: Rollo

Originally posted by: deadseasquirrel

Originally posted by: apoppin
edit: OMG, i just agreed with Rollo :Q

Live pictures from hell.

:laugh:

ROFL!

 

[Drayvn]

Originally posted by: 5150Joker
Looks like XTX are cherry picked cores and that's probably why ATi is charging $100 extra for them. I got the itch to OC the new XTX I bought last night and it clocks to 750/900 without breaking a sweat. I think I can easily push the core to 800 mhz or higher but want to wait until the AS5 settles in and my ramsinks arrive. The XT I had on the other hand was barely pushing 720 core and 840 memory before artifacting. I've been reading xtremesystems forums and with the exception of a few people, it seems the XTX cards do clock higher than the XT. So despite having the same ram, most of the XT's don't clock as well as the XTX do. I suppose if you are an overclocker, that extra $100 for the XTX may be worth it.

I dont know about my assumption is its correct or not but no one has brought it up. Here goes.

If the XTX cores are cherry picked then either the manufacturing process or design are being overloaded? Or something like that?

Well if they get either problem working correctly wouldnt that mean around the same time the G71 will come out or possibly slightly later wouldnt they bring out a higher clocked model?

 

[5150Joker]

Originally posted by: RobertR1
Joker,

Would you give us some FEAR and even 3dmark06 benchmarks, before and after the OC? I'd like to know how well it scales with raised clocks and extra bandwidth.

Don't have FEAR installed but I'll see if I can install it tonight. 3DMark06 I'll give a run later but keep in mind my score will suck because of the CPU.

 

[5150Joker]

Originally posted by: Pete
Aren't the stock voltages on the XTX higher than on the XT? That might explain OCing differences.

AFAIK the voltages for both cards are the same.

 

[DeathReborn]

Originally posted by: deadseasquirrel

Originally posted by: apoppin
edit: OMG, i just agreed with Rollo :Q

Live pictures from hell.

This one has a bit more effect I think: hell

 

[Killrose]

Originally posted by: 5150Joker
Looks like XTX are cherry picked cores and that's probably why ATi is charging $100 extra for them. I got the itch to OC the new XTX I bought last night and it clocks to 750/900 without breaking a sweat. I think I can easily push the core to 800 mhz or higher but want to wait until the AS5 settles in and my ramsinks arrive. The XT I had on the other hand was barely pushing 720 core and 840 memory before artifacting. I've been reading xtremesystems forums and with the exception of a few people, it seems the XTX cards do clock higher than the XT. So despite having the same ram, most of the XT's don't clock as well as the XTX do. I suppose if you are an overclocker, that extra $100 for the XTX may be worth it.

I dont doubt that at all, but then again there is alway's the fact that every core will overclock differently as is well known with CPU's. And AMD, Intel get more money for their CPU's that perform better so why not?

 

[Elfear]

Here are a few FEAR benchies I did last night for those interested.

X1900XTX@stock 1920x1200 2AA/8AF SS off and everything else max

Min: 31
Avg: 53
Max: 121

0% below 25fps
32% between 25fps and 40fps
68% above 40fps

X1900XTX@722/872 same settings as above

Min: 33
Avg: 58
Max: 136

0% below 25fps
28% between 25fps and 40fps
72% above 40fps

This is what I got with my 7800GTs:

7800GTs@stock same settings as above

Min: 23
Avg: 50
Max: 183

3% below 25fps
7% between 25fps and 40fps
90% above 40fps


7800GTs@550/1300 same settings as above

Min: 48
Avg: 65
Max: 208

For the people who don't believe ocing is worth it, see the above results. The 7800GTs and Opteron chip combo I was running were pretty dang close to the performance of SLI'd 512MB 7800GTXs and an FX-57 (Opteron was actually faster) yet instead of spending $2500 on my rig (if you could even find the 512MB 7800GTXs in stock at $750) I only spent $850. That's almost a 1/3 the cost. Easy for me to see why people oc.

 

[Dman877]

To the OP: they call it speed binning, not cherry picking and it's been part of computer hardware manufacturing forever.

 

[M0RPH]

Originally posted by: Dman877
To the OP: they call it speed binning, not cherry picking and it's been part of computer hardware manufacturing forever.

Yeah, the words 'cherry picked' don't really apply here, but I knew what he meant. Anyone who thinks that 'cherry picked' is used in a negative way here missed the point.

 

[5150Joker]

Originally posted by: M0RPH

Originally posted by: Dman877
To the OP: they call it speed binning, not cherry picking and it's been part of computer hardware manufacturing forever.

Yeah, the words 'cherry picked' don't really apply here, but I knew what he meant. Anyone who thinks that 'cherry picked' is used in a negative way here missed the point.

Yeah that's essentially what I was getting at. The reason I even brought it up is because a lot of people (myself included) and websites initially dismissed the XTX as not being worth the extra $100 compared to the XT due to the slight clock advantage. That assertion is still true if you don't plan to overclock but if you are an overclocker, the XTX is well worth it.

 

[Pete]

Originally posted by: 5150Joker
AFAIK the voltages for both cards are the same.

Hmmm. I thought I read that somewhere, but maybe I'm thinking of the X1800XL and XT? I guess a X1900XT owner with RivaTuner could compare their core voltage to the X1900XTX core's 1.45V. It's surprising to see the XTX's extra 25MHz core and RAM clocks cause a 20W raise in power draw with equal voltages, especially if that's on top of just 120W for the XT. A 15% increase in power draw for ~4% clock increases? Maybe TR's #s are skewed due to engineering samples.

 

[M0RPH]

Pete, I also remember reading that from someone here in the forum. Would be nice if someone could confirm it for sure either way.

 

[Piuc2020]

Well the reason people overclock top end GPUs is the very same reason why people even buy top end GPUs in the first place, to get the best possible performance ever, why buy a X1900XTX when you can play on high all fine with a 7800GT?

 

[Wall7486]

It depends what you mean by high all fine. 7800GT is a great card, but pair it up with a monitor like the Dell 3007 and playing everything at its native resolution might be choppy. People buy these x1900Xtx/ 7800GTX 512MB so they can play on 1900x1200 res at the very least with full AA/AF on, at least I hope that's what they do. I'd love to get my hands on one of those X1900xtx, maybe when more $$ is coming in.

 

[5150Joker]

Originally posted by: Elfear
Here are a few FEAR benchies I did last night for those interested.

X1900XTX@stock 1920x1200 2AA/8AF SS off and everything else max

Min: 31
Avg: 53
Max: 121

0% below 25fps
32% between 25fps and 40fps
68% above 40fps

X1900XTX@722/872 same settings as above

Min: 33
Avg: 58
Max: 136

0% below 25fps
28% between 25fps and 40fps
72% above 40fps

This is what I got with my 7800GTs:

Ok ran the benchmark at 740/830, it won't go higher than that in FEAR without locking up although it runs through 3dmark05 at 750/900 so could be FEAR is stressing the card more and creating more heat. The ram doesn't have anything cooling it so it gets scorching hot.

Settings used are quality settings in CCC, no SS, no angle independent AF, 1680x1050 4xAA/16xAF with everything in the game options maxed out:

Click

and this is with the same settings above but the card at default speed:

Default Speed

I'm fairly certain once I get some ramsinks and my X2 4400+ ever gets shipped from Dell the scores will definitely go up. Right now I'm limited by my 3000+ winchester@2.5 ghz. Also, the ATi OC tool reads the core voltage as 1.2v rather than 1.45v that digit-life reported with their engineering sample.

 

[firewolfsm]

 

[ElFenix]

so by buying a card that does 25 mhz more on the core and 50 mhz more on the ram, you've gotten it overclocked by an additional 5 mhz and 10 mhz, respectively?

 

[5150Joker]

Originally posted by: ElFenix
so by buying a card that does 25 mhz more on the core and 50 mhz more on the ram, you've gotten it overclocked by an additional 5 mhz and 10 mhz, respectively?

Huh? The XT was doing those speeds in 3dmark05 vs 750/900 for the XTX. Apparently FEAR is stressing the card a lot more than 3dmark05 does during its test. So for FEAR the XTX won't OC above 740/830 w/out artifacting. I'm guessing the XT would have likewise lost as much OC headroom thus bringing it down back to the 700/780-800 range. I don't have the XT with me anymore so I can't confirm this but someone else w/an XT should be able to.

 

[OvErHeAtInG]

Originally posted by: Pete

Originally posted by: 5150Joker
AFAIK the voltages for both cards are the same.

Hmmm. I thought I read that somewhere, but maybe I'm thinking of the X1800XL and XT? I guess a X1900XT owner with RivaTuner could compare their core voltage to the X1900XTX core's 1.45V. It's surprising to see the XTX's extra 25MHz core and RAM clocks cause a 20W raise in power draw with equal voltages, especially if that's on top of just 120W for the XT. A 15% increase in power draw for ~4% clock increases? Maybe TR's #s are skewed due to engineering samples.

No I think AT had similar numbers. An XTX pulls more watts under load than a GTX512, while the XT pulls less. IIRC.

 

[WhoBeDaPlaya]

Originally posted by: deadseasquirrel
Live pictures from hell.

ROTFLMAO!

 

[Sentry2]

Joker you are nuts for running that card without ramsinks on it.....and then oc'ing without them too....:Q