Shamino's Full Review of The X1950XTX

As seen through this link.

http://www.vr-zone.com/index.php?i=3885&s=1

People will be much surprised to find out the X1950 actually uses 36W MORE than the current X1900XTX despite the rumours it would consume less power. It was believed this would be so due to a simpler PCB layout and one less mosfet on the X1950 over the current X1900 but I guess that was to be taken with a grain of salt.

Either way I wont ruin much more of the review as I would like you guys to enjoy it as much as I did while going through it. I am happy to say performance is much above the current X1900 cards and overclocked it gets even better.

Enjoy!

nicely written review, the card looks really promising.
Some improvements with the oc tools and it will fly.
Wonders me that this card outperforms a GX2 in games but it also uses more wattage :eek:

Wait for nda to expire on full oc results ;)

Perkam

It is not about OC, but Power consumption... 70W more in CrossFire @ default ...:mad: it is crazy ...

Quote:

---

Originally Posted by perkam

Wait for nda to expire on full oc results ;)

Perkam

---

yes indeed, shamino using only air cooling for overclocking ... I didn't think so :D

750watts should be enough to run one x1950xtx and also a future x1950xtx crossfire setup right?

Awesome, exactly what I wanted to see. Better performance in games at the exact resolution I play at! Can't wait to get one!!:toast:

Quote:

---

Originally Posted by rpg711

750watts should be enough to run one x1950xtx and also a future x1950xtx crossfire setup right?

---

The 750W Silverstone ZF can achieve 900W continous max output.

Perkam

single X1950XTX vs 7950GX2

the X1950XTX lost in the 3Dmark test (lower score than 7950 in 06

but It won all the game test:cool:

X1950XTX 512MB 256bit 0.85ns GDDR4 @ 650/2000
$449

7950GX2 1GB 512bit 1.1ns GDDR3 @ 500/1200
$599-$699

the X1950XTX will be the fastest single card if it OC to this monster clock
http://service.futuremark.com/compare?3dm05=1911325

the 80nm ATi R580+ core and Samsung 0.85ns GDDR4 maybe rush to 1100/2300


http://news.mydrivers.com/pages/imag...2606_91549.jpg

and now the mainstream video card benchmark is

3Dmark03 & 05 benchmark (ranked by 05)

AMD Athlon FX55 , all in stock 1024x768 noAA noAF

card name --3Dmark03--3dmark05---clock---
X1950XTX -----24764-----13231 @ 650/2000
7950GX2-------29342-----12886 @ 500/1200
X1900XTX -----18492-----11065 @ 650/1550
7900GTX ------21412-----10522 @ 650/1600
X1900XT ------17598------9959 @ 625/1450
7300GT SLI----20364------9516 @ 550/1600
7800GTX512 ---19959------9509 @ 550/1700
X1900GT ------16044------9359 @ 575/1200
X1800XT ------13951------9203 @ 625/1500
7900GT -------16616------7930 @ 450/1320
7800GTX ------16373------7789 @ 430/1200
X1800GTO3-----13236------7356 @ 500/1000
X1800XL ------13522------7122 @ 500/1000
7800GT -------14589------6885 @ 400/1000
X1800GTO -----12636------6785 @ 500/1000
X850XT-PE-----13197------6574 @ 540/1180
X850XT -------12702------6351 @ 520/1080
X800XT-PE ----12484------6285 @ 520/1120
X800XT -------11842------5953 @ 500/1000
7600GT -------13163------5935 @ 560/1400
7800GS -------13280------5893 @ 375/1200
6800Ultra ----12985------5843 @ 400/1100
X850PRO ------10059------5426 @ 507/1040
X1600XT -------9034------5228 @ 590/1380
7300GT -------12187------5174 @ 550/1600
6800GS -------11134------5057 @ 425/1000
X800XL -------10192------4986 @ 400/1000
6800GT -------10865------4855 @ 350/1000
X800GTO3 -----10087------4754 @ 400/980
X800PRO -------9984------4733 @ 475/900
7600GS -------11846------4721 @ 500/1400
X1600PRO ------8277------4532 @ 575/1300
X800 ----------9306------4283 @ 400/700
X800GT --------8904------4052 @ 475/980
6800NU --------8354------3759 @ 325/700
6600GT --------8152------3427 @ 500/1000
9800XT --------6954------3195 @ 412/730
6800LE --------6752------2957 @ 325/700
5950Ultra -----6541------1453 @ 475/950

3d mark is just something to bench... and brag about
actual game performance matters most
it seems like fear is the only game that utilized the 7950gx2 to its most performance

Now that was one damn fine review, good job Sham :toast:

I love that cooler, quieter and the mem sink is just sick. Then I read what the projected retail cost is :slobber: :slobber: :D ATI seems to have really dropped the hammer with this card, I'll take a little more power consumption for everything else this beauty has to offer anyday.

im still skeptical on the so called launch price. I feel it will be much higher than predicted but only time will tell.

Quote:

---

Originally Posted by G H Z

Now that was one damn fine review, good job Sham :toast:

I love that cooler, quieter and the mem sink is just sick. Then I read what the projected retail cost is :slobber: :slobber: :D ATI seems to have really dropped the hammer with this card, I'll take a little more power consumption for everything else this beauty has to offer anyday.

---

but will ya buy one or wait for the next:D

I love the independant solid copper ramsink.

that will be extremely useful for watercooling, so you can just use a Maze 4 or something, and not need to adhere heatsinks to anything.

Definately buying, only drawback is whether or not you'll get one with no bug and big core clocks ;)

Quote:

---

Originally Posted by milkcafe

single X1950XTX vs 7950GX2

the X1950XTX lost in the 3Dmark test

but It won all the game test:cool:

X1950XTX 512MB 256bit 0.85ns GDDR4 @ 650/2000
$449

7950GX2 1GB 512bit 1.1ns GDDR3 @ 500/1200
$599-$699

the X1950XTX will be the fastest single card if it OC to this monster clock

the 80nm ATi R580+ core and Samsung 0.85ns GDDR4 maybe rush to 1100/2300

---

don't always trust benchmarks released from one side

Quote:

---

Originally Posted by cantankerous

im still skeptical on the so called launch price. I feel it will be much higher than predicted but only time will tell.

---

the x1900xtx costs so much because the same reason the 7900gtx costs so much, because people will buy them for that much
the x1950xtx costs so little because only the hard core gaming community will care about the extra 3 fps...
the enthusiasts mabee will just get one to brag about
all others will probably stay with their x1900xt's

HOT damn!! and i do mean HOT...

and here i thought i was having a hard time keeping my 1900s' cool... now were going to have even MORE BTU's to move :eek: :eek:


are these single card only or are CrossFire editions just around the corner?


and here i just got CF working on the Conroe rig last week :rolleyes:


PS GREAT review !!!!

Thanks... this card surprised me in some areas, first when i expected it to draw lesser power it goes the other way around, then when i just expected alightly better numbers from 19xtx it made me rerun the tests again cos quite frankly i was quite skpetical the difference was like this. I know how it may look, bought review, but i wish it was bought cos if so now i'd have a x1950 cross-fire to play with :D
the card was passed back already, was like some rare gem i had only a short time with :nono:
i also woulda wished to test for cold bug, thats the whole world of importance for guys like me and maybe u ;)

crossfire editions are comming as well
it wont be as hot because it has better cooling than any other card on the market so far
and it definately wont be hot for me with a 265cfm fan intaking

sham....... dooooooooooooooooood.................

whats up with that thumbnail in the "back of the core temp reading" pic?
are you a werewolf? :)

stock coolers stay on for about 1-2 days till the bugs are worked out in a rig then its time to move on...


(obviously, these are 1900's NOT 1950's!!)

http://img528.imageshack.us/img528/9...0285iu7.th.jpg http://img147.imageshack.us/img147/3...04541lb.th.jpg

(ViperJohn Ramsinks prohibit me going back to stock...so i ghetto-ed up some air cooling :rolleyes: )



.

36w more power, and that is before oc. Looks like my 600w powerstream is yet again pushed along a little more. I hope it will keep up.

my silverstone zeus 750 watt continuous peak 900watt will keep up :D

yeah, SHAM is my favorite review author... only thing I'm worried about is the "clock for clock" comparison with the x1900xtx...

Dayum, it beats the gx2 that bad?

There has to be something wrong here..

Im on the fone right now with EVGA stepup trying to find my gx2, now I read this :(

He comments that he thinks oblivian may be cpu limited, which begs the obvious, why didn't he overclock the conroe to find out? Ugh. Even an easily applied mild overclock would have given the answer rather than leaving it as a what if.

He didn't have that card for very long, I'm surprised they even let him take the cooler off. Then again, maybe they didn't :D

Quote:

---

Originally Posted by keiths

He comments that he thinks oblivian may be cpu limited, which begs the obvious, why didn't he overclock the conroe to find out? Ugh. Even an easily applied mild overclock would have given the answer rather than leaving it as a what if.

---

thats because he didnt have the card long enough
hes trying to get it back or soemthing...
he only had time to write the review
it stinks how u cant just get a x1950xtx ES like u can get a Conroe ES

The link goes to VR-Zone, but to no page in particular. There's no title, and it says it was written in 1970. :shrug:

thats because they took it down becuase it wasnt full in info

Quote:

---

Originally Posted by rpg711

thats because they took it down becuase it wasnt full in info

---

not really, was "forced to" pull it by "contact"

ok :(
that stinks
well... enough people read it alredy... atleast i did :D
i remember something about it not being full in info... that might be hardware zone

Its okay Sham, pulling review is fine I'm sure a few people cache'd it.

Can anyone post the benchmarks from this review? I've been at work all day and missed it!

(1st page)

Quote:

---

Originally Posted by LardArse

not really, was "forced to" pull it by "contact"

---

What?? Red or Green Team ;)

Quote:

---

Originally Posted by G H Z

What?? Red or Green Team ;)

---

It´s Green or Green team nowadays :p:

Quote:

---

Originally Posted by LardArse

Thanks... this card surprised me in some areas, first when i expected it to draw lesser power it goes the other way around, then when i just expected alightly better numbers from 19xtx it made me rerun the tests again cos quite frankly i was quite skpetical the difference was like this. I know how it may look, bought review, but i wish it was bought cos if so now i'd have a x1950 cross-fire to play with :D
the card was passed back already, was like some rare gem i had only a short time with :nono:
i also woulda wished to test for cold bug, thats the whole world of importance for guys like me and maybe u ;)

---

mybe this is why it runs hot

ready for the big freez

ok forget the game benchmark numbers for the time being, I got flamed pretty bad all around by readers, I gotta figure out what went wrong with my 7950GX2 runs.

Since the date i can publish the review is pushed back quite a bit so i'll redo it and redo the setup.

the rest is still pretty relevant though

Quote:

---

ok forget the game benchmark numbers for the time being, I got flamed pretty bad all around by readers, I gotta figure out what went wrong with my 7950GX2 runs.

---

Are you saying you're wrong or the readers are wrong?

For my part, i recently had to set up a 7950gx2 system. First time round, i installed the 91.31's, and Prey performance was shocking. I then upgraded to the 91.45's, which improved matters, but not much. 3dmark03 scored only 22,500. Tried driver cleaning and everything, but couldn't improve performance at all.
We were totally puzzled, but we reformatted, installed the 91.45's from the off, and everything was happy again.

That's probably a mistake far too basic for you, but worth a try?

Sham, damn, I thought it was under NDA:..
But still love your work :D

Anyhow, you said something about coldbug, any news on that front?
*Put your cascade on the sucker*

EDIT:
Ohh, another important thing... THe holes, same location as X1900XTX, right?? - I kind'a hope so.. For the watercooling's sake :)

Quote:

---

Originally Posted by M.Beier

Sham, damn, I thought it was under NDA:..
But still love your work :D

Anyhow, you said something about coldbug, any news on that front?
*Put your cascade on the sucker*

EDIT:
Ohh, another important thing... THe holes, same location as X1900XTX, right?? - I kind'a hope so.. For the watercooling's sake :)

---

cold bug until i get a card i can keep no way i can try, there's no way they're gonna let me off do insulative job on this card i hardly have much time with
but if u want my gut feel its gonna be the same with x19s, cold bug, but cascadable with vlots and clocks and 3ding to keep warm.
yep same holes

Quote:

---

Originally Posted by don_vercetti

Are you saying you're wrong or the readers are wrong?

For my part, i recently had to set up a 7950gx2 system. First time round, i installed the 91.31's, and Prey performance was shocking. I then upgraded to the 91.45's, which improved matters, but not much. 3dmark03 scored only 22,500. Tried driver cleaning and everything, but couldn't improve performance at all.
We were totally puzzled, but we reformatted, installed the 91.45's from the off, and everything was happy again.

That's probably a mistake far too basic for you, but worth a try?

---

heheh u tryin to trap me here with that question :P
yes thats what i'll do, always had nop problems without reinstaling drivers and such but this time seems like it needs something more to flush out.,

Hmm, same temperature as X1900XTX, that means, thumbs up for -25deg water in waterchilling :D

Great

Shamino, did you tried, figured out already mod, for countering coldbug, like hipro did with 1900XTX?
You said in review that that part of pcb is different.

Any one else having problem seeing anything in that article? It looks like this both in IE and FF.

http://img228.imageshack.us/my.php?image=cantseead7.jpg

It's deleted from vr-zone, but you can find it elsewhere :)

Quote:

---

Originally Posted by Kriller

It's deleted from vr-zone, but you can find it elsewhere :)

---

Tried finding it in googles cach, but no luck.

Care to post a link, please?

http://resources.vr-zone.com.sg/Shamino/1950/

Here are some benches from HardwareZone before they pulled it.

http://www.hardwarezone.com/img/data...0/RiddickB.gif
http://www.hardwarezone.com/img/data...6/2020/Q4b.gif
http://www.hardwarezone.com/img/data...06/2020/SC.gif
http://www.hardwarezone.com/img/data...2020/FearA.gif
http://www.hardwarezone.com/img/data...2020/FearB.gif
http://www.hardwarezone.com/img/data.../3dmark05a.gif
http://www.hardwarezone.com/img/data.../3dmark05b.gif
http://www.hardwarezone.com/img/data...0/3dmark06.gif
http://www.hardwarezone.com/img/data.../2020/Temp.gif

I don't know why this review looks so much slower performance wise for the 1950XTX than vr-zone.

Maybe they realized they screwed up their benchmarks and are redoing them right now.(maybe vise-versa :D)

Anyway these are the pics of their results BEFORE they pulled their review article, enjoy.

Quote:

---

Originally Posted by Your_Boss

http://resources.vr-zone.com.sg/Shamino/1950/

---

Thank You

Quote:

---

Originally Posted by LardArse

ok forget the game benchmark numbers for the time being, I got flamed pretty bad all around by readers, I gotta figure out what went wrong with my 7950GX2 runs.

Since the date i can publish the review is pushed back quite a bit so i'll redo it and redo the setup.

the rest is still pretty relevant though

---

Thats BS man. If you can run all the 3DMarks and get a higher score, the games should work just the same.

Quote:

---

Originally Posted by ahmad

Thats BS man. If you can run all the 3DMarks and get a higher score, the games should work just the same.

---

I think games and GFX benchies are two diff things. In a game you'll have alot of area's that require little of the card but when alot of action is on the card its stressing it. Benchies are full load high res trying to render large maps and intricate designs. When your getting shot and so much comotion is going on in game I think its a bit different from 3dmark my self :)

thanks ever so much for those graphs. I missed the original article and was searching everywhere for the results it produced.

EDIT... just gave a quick glance at the results and I must say I am dissapointed, especially after reading the VR-Zone results which painted a much prettier picture of the card. I am upset to see the 3DMark scores actually go down while any gains seen elsewhere were extremely minimal. I am glad to see the temps are a little lower, but without info on system tested on, fan speed setting, overclock (if any), as well as ambient temps during testing, makes this review hard to go by as well. Shamino just hinted the card may be pushed back until the middle of September before release, so that should give more time for other reviews, hopefully end users, to put in their commends and results for comparison purposes. I was stoked about getting this card and selling my current but perhaps it really isn't worth it after all.

Quote:

---

Originally Posted by Your_Boss

Here are some benches from HardwareZone before they pulled it.
..............
I don't know why this review looks so much slower performance wise for the 1950XTX than vr-zone.

Maybe they realized they screwed up their benchmarks and are redoing them right now.(maybe vise-versa :D)

Anyway these are the pics of their results BEFORE they pulled their review article, enjoy.

---

Thats because:

1: Crappy AF settings. These are high end cards that can handle it
2: Chronicles, Quake4, FEAR (was even done with no AA) are all games we know that Nvidia is better at (also, FEAR takes advantage of the raw power of 2 cards). Who cares about OpenGL performance?

Sad. Why dont they show some HL2 Episode 1, or serious sam, or farcry. Hardwarezone goes on the list of sites not believe anything from :nono:

Quote:

---

Originally Posted by ahmad

Thats BS man. If you can run all the 3DMarks and get a higher score, the games should work just the same.

---

u dont have a clue what ur talking about
ddr4 memory has higher latencies, hence the lower 3d mark scores
latencies matter only in benchies and software stuff
latencies matter none in games, where the x1950xtx's memory bandwidth is used to higher potential, the x1950xtx's memory bandwidth is used more efficiently than the 7950gx2... and the memory clock also matters alot in games

If they pulled it, they screwed up the benchmarks bottom line, there is no mystery there.

Actually Hardware Zones look pretty accurate all around compared to other reviews....except for the new X1950XTX results of course. The GX2 and X1900XT results look right on the money.

Quote:

---

Originally Posted by LardArse

ok forget the game benchmark numbers for the time being, I got flamed pretty bad all around by readers, I gotta figure out what went wrong with my 7950GX2 runs.

Since the date i can publish the review is pushed back quite a bit so i'll redo it and redo the setup.

the rest is still pretty relevant though

---

Just because you got flamed by people who didn't believe your results doesn't mean you should scrap your findings as irrelevant. Some do believe your results to be correct.

Time will tell whether you were acurate or not. Until then, don't concern yourself with what other angry 7950GX2 owners are yelling at you for. You did what you could as accurately as you could with the time frame you were given. Thanks.

Quote:

---

Originally Posted by Vengure

I think games and GFX benchies are two diff things. In a game you'll have alot of area's that require little of the card but when alot of action is on the card its stressing it. Benchies are full load high res trying to render large maps and intricate designs. When your getting shot and so much comotion is going on in game I think its a bit different from 3dmark my self :)

---

No, you have it backwards. Since games are run at high res and AA + AF the video card is getting stressed hence the small differances in perf. between cpu's and other small things in gaming environments. Only the gpu matters (for the most part) in gaming environments. Since benchmarks are run at lower res and settings and are made to evaluate the entire system's performance cpu and other sublities such as ram timings etc. play a much bigger role in the results.

I think they were pulled after people had words rather than any questions on the results.

Regards

Andy

No offence to you shamino, we all know your top in the overclocking community and well respected bro..

Im glad to see that it was pulled ^^^ I actually soaked my pillow in tears after accepting my stepup from evga thinking I was finally gonna have a top performing card, and then read your article and got shot down hard..Hopefully after they allow you another shot at the card itll be a different story..

Quote:

---

Originally Posted by mdzcpa

Actually Hardware Zones look pretty accurate all around compared to other reviews....except for the new X1950XTX results of course. The GX2 and X1900XT results look right on the money.

---

I think you're right Mike. Will be interesting to see how it looks after a few other reviews / tests have been run using some other drivers on both sides. I'd expect the Xfire with the X1950 to compete with the 7950GX2, but not the single card. I'm not a fanboy for either camp, if one is substantially faster than the other for the same money - I'll buy it. I also consider the noise and heat levels too - and the NVidia options have been better in those two areas lately. I hope the X1950 changes that.

Does LardArse = Shamino?

Quote:

---

Originally Posted by Philosophy

No offence to you shamino, we all know your top in the overclocking community and well respected bro..

Im glad to see that it was pulled ^^^ I actually soaked my pillow in tears after accepting my stepup from evga thinking I was finally gonna have a top performing card, and then read your article and got shot down hard..Hopefully after they allow you another shot at the card itll be a different story..

---

:confused:

what do you mean?

Quote:

---

Originally Posted by kenofstephen

:confused:

what do you mean?

---

he means that he got his 7950GX2 from evga and then he finds according to this review that X1950XTX AND X1900XTX is better than his card which is not true

will there be a need for a master card for cf? or can you just use two x1950xtx's?

Quote:

---

Originally Posted by Origin_Unknown

will there be a need for a master card for cf? or can you just use two x1950xtx's?

---

Yeah you will need a X1950 master card. It will cost the same as the slave and have the exact same specs.

That's a pleasant change. Do they all clock +- equally as well?

Quote:

---

Originally Posted by Kinc

Yeah you will need a X1950 master card. It will cost the same as the slave and have the exact same specs.

---

hmm, thought they were getting rid of that and having it done internally (sli style)

After reading this review (Before it was pulled) it seems quite obvious that the 7950, was not running both cards, but only a single card was functioning properly. (In the gaming tests)

Whether or not it was software or hardware related i'm sure Shamino will figure it out and run a new test.

Shamino did not pull out the review because the benchmarks where not correct, he pulled it out because the red guys asked him to do it.

If Peter says that are the results then those are the results, bottom line.

Hey take it easy now:D
I am not questioning his integrity.;)

He knows there was something wrong with the 7950 results and has said as much. :cool:

Quote:

---

Originally Posted by don_vercetti

That's a pleasant change. Do they all clock +- equally as well?

---

all cards clock differently, because all cores are different
if u get two cards from the same batch, one may overclock to 900/1500 and one might not at all... silicon is weird

Quote:

---

Originally Posted by rpg711

all cards clock differently, because all cores are different
if u get two cards from the same batch, one may overclock to 900/1500 and one might not at all... silicon is weird

---

Helpful answer. Read my post again, think about it, and then try and come up with a constructive reply.

I said +- well. I know all silicone is not equal, everyone does. It's just that the X1900 Crossfire was based off the X1900XT rather than XTX, so it didn't clock as well as an X1900 Crossfire might have done had it been based off XTX, hence limiting the OC potential of Crossfire somewhat. However if the X1950 Crossfire clocks as well as the X1950 XTX...then we're very happy!

ohhh i c what u mean
the x1950 crossfire should be called x1950xtx crossfire edition because it is based off of the x1950xtx... just has a crossfire stuff thingy dohicky on it :p:

lets hope so eh.

maybe this is me but why is the article saying the date is "Date: January 1, 1970, 3:30 pm " and its not showing any info on it at all...

it was pulled... if u read the last page of stuffs, things, and thingies

Quote:

---

Originally Posted by don_vercetti

Helpful answer. Read my post again, think about it, and then try and come up with a constructive reply.

I said +- well. I know all silicone is not equal, everyone does. It's just that the X1900 Crossfire was based off the X1900XT rather than XTX, so it didn't clock as well as an X1900 Crossfire might have done had it been based off XTX, hence limiting the OC potential of Crossfire somewhat. However if the X1950 Crossfire clocks as well as the X1950 XTX...then we're very happy!

---

Silicone is for boobs, Silicon is for CPUs :p:

http://enthusiast.hardocp.com/articl...VudGh1c2lhc3Q=

"It was quite shocking in some cases to see the Sapphire TOXIC X1900 XTX actually perform faster than the BFGTech GeForce 7950 GX2."

x1950 > 7950GX2 doesn't surprise me at all.

Just imagine had they compared most games with 4xAA for both cards, rather than comparing 4xAA GX2 with 6xAA for X1900xtx, as well as setting the GX2's quality settings in the driver to match ATI's

Quote:

---

Originally Posted by JuanFlaiter

Silicone is for boobs, Silicon is for CPUs :p:

---

What's an E between friends, eh?

Why did they have to go and push the launch date back further. I don't know why I want this card so bad but I do. I hope it over clocks as well as the x1900xtx.

Quote:

---

Originally Posted by Daveb2012

Why did they have to go and push the launch date back further. I don't know why I want this card so bad but I do. I hope it over clocks as well as the x1900xtx.

---

aparently it's for supply. But it seems they'll launch on the 23rd, but without availability until the second week of september.

Hi

Can anybody tell me how much watts eats X1900XTX in max stress and how much X1950 (according to article) ?

That article doesn't work for me :( .

Regards
Peter

Quote:

---

Originally Posted by piotres

Hi

Can anybody tell me how much watts eats X1900XTX in max stress and how much X1950 (according to article) ?

That article doesn't work for me :( .

Regards
Peter

---

x1900xtx here

http://www.xbitlabs.com/articles/vid...wer-noise.html