![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
First Review of X1950XTX from VR-Zone
- Thread starter Overgloc
- Start date
- Status
Spriggan_X(ò_ó)
n00b
- Joined
- May 29, 2006
- Messages
- 35
omg, so god-like, me wants crossfire, bow to ati xD.
AltTabbins
Fully [H]
- Joined
- Jul 29, 2005
- Messages
- 20,387
nice cooler.. (edited)
tornadotsunamilife
2[H]4U
- Joined
- Jun 26, 2004
- Messages
- 2,492
I thought the close-up pictures were of the card
AltTabbins
Fully [H]
- Joined
- Jul 29, 2005
- Messages
- 20,387
tornadotsunamilife said:
whoops.. EDITED
tornadotsunamilife
2[H]4U
- Joined
- Jun 26, 2004
- Messages
- 2,492
Well either way I'm not convinced with it's performance
Asian Dub Foundation
Supreme [H]ardness
- Joined
- Aug 1, 2005
- Messages
- 7,940
sexy cooler though. all vga cooler should exhaust heat through the back
skittzle said:QFT
This review shows the x1900xtx beating the 7950gx2 which goes against about everything I have read. Maybe they forgot to enable the other gpu on the 7950
It's weird but a bunch of recent reviews have been showing ATI doing great.
I'm thinking the newest drivers may have some really big performance increases.
Anyway thinking of it from a larger perspective, one can draw some odd conclusions.
The Nvidia solution is really TWO cards. Why is it being benched against one?
Nvidia die is about 1/2 size, but with two, it's about as big as ATI..but it still loses.
Nvidia got a little too clever for themselves, whenever you SLI cards you lose a lot of efficiency..in this case, had they made one GPU with the die size of 7950GX2, they'd be killing, instead by taping two together, they are making themselves a lot less efficient
Technically speaking, yes, it is two cards. However, it comes in a single box, it occupies a single PCIe slot, occupies the same space as any other "dual-slot" card (including the X1950), and is sold and labelled as a single, all-encompassing product. Whether or not it uses two physical GPUs or thirty-seven is somewhat irrelevant. Yes, it does operate using SLi, and yes, SLi is not highly effecient, but it is still a "single card". You can't take it apart and use one (at least I don't believe you can...), which makes it a single "graphics card product", if you wish.Sharky974 said:
I suppose, by your standards, that it is also heresy to compare a CrossFire/SLi solution to a single card in a review? I for one am glad that this isn't the case, as we'd all have to do a hell of a lot of cross-referencing to try and figure out what SLi and CrossFire really do for us.
It should always be up to the reader to understand how a product operates and what kind of limitations are inherent in the product. I can compare a Lian Li computer case to a banana in a review if I so choose; it is up to the reader to decide if the banana is going to be an acceptable enclosure for his new rig.
phide said:
They use HQ for NV, not Q. The way it should be reviewed. Most other reviews use Q, which has worse IQ than ATi, HQ gets it closer to ATi's quality, and more apples to apples. HQ takes a rather large performance hit, over Q. Which is why NV tells reviewers to use Q. The default quality setting for NV is Q, or 3/4 of the quality settings. ATi's defaults to the highest setting, 4/4.
edit, here is a link to the review, since the original has been pulled; http://resources.vr-zone.com.sg/Shamino/1950/
It doesnt have everything, but has most of it.
the reviewer himself said something was wrong, so wait for the fixed benchmarks before you come to any conclusions. I know that might be very hard for you fallguy.
That was his last post, so he still felt something was wrong
BTW being an nV fan, also I'm not, might want to check back the bug in FEAR and Oblivion has been fixed, and this is why a few members here thought I was biased because I said it was a bug, and now that is now true, when a game doesn't scale like other games something is wrong with the drivers, nothing to do with the hardware.
I know how fast the x1950xtx is and was when the memory clocks were released, there is no maricle 50% boost in performance over the x1900xtx. The gf7950 gx2 never lost to the x1900xtx. You know that very well.
That was his last post, so he still felt something was wrong
BTW being an nV fan, also I'm not, might want to check back the bug in FEAR and Oblivion has been fixed, and this is why a few members here thought I was biased because I said it was a bug, and now that is now true, when a game doesn't scale like other games something is wrong with the drivers, nothing to do with the hardware.
I know how fast the x1950xtx is and was when the memory clocks were released, there is no maricle 50% boost in performance over the x1900xtx. The gf7950 gx2 never lost to the x1900xtx. You know that very well.
He said what "may" go wrong. The Fear/GX2 numbers are the only ones in question, ass far as I can tell. Even though it shows the X1900XTX at 32fps, the GX2 at 34fps, and the X1950XTX at 40fps, without AA. Seems pretty reasonable to me.
And I have not come to any conclusions, yet. Dont assume you know all, as has been shown in the past, you dont.
And I have not come to any conclusions, yet. Dont assume you know all, as has been shown in the past, you dont.
fallguy said:
I'll bet you 100 bucks if ya want
, I don't know all but I do know my math is always right Having the hardware benchmarks doesn't mean sh*t if you can understand what it all means and possible errors. You do jump to conclusions maytimes sepecially when it concerns anything that might be positive to nV, actually almost always.
razor1 said:I'll bet you 100 bucks if ya want
Having the hardware benchmarks doesn't mean sh*t if you can understand what it all means and possible errors. You do jump to conclusions maytimes sepecially when it concerns anything that might be positive to nV, actually almost always.
Wrong. I dont even read vrzones reviews. I have a select few I like. Nobodys maths is always right, another instance of you being wrong...
I rarely jump to conclusions, another instance of you being wrong. If one, or a few of my fav hardware sites show the same thing, I believe it. I dont believe the Inq crap, or any other rumors.
Yes the GX2 numbers look low for Fear, especially because its typically the "poster chold" for SLI, it generally gets a huge boost. However, HQ takes a rather large hit from Q. And HQ is how all reviews should be done, but that wont happen. Perhaps you can look at older reviews and compare numbers, I dont care to. But as I said, thats the only numbers being questioned. So calling his whole review void because of that, is pretty silly. If you dont agree, then on well. Im done discussing it.
X1900XT oc'ed to X1900XTX -> 7950GX2 was a pretty big performance increase for me.
X1900XTX -> X1950XTX isn't a big jump. Faster memory, that's it. It's not enough to make a huge performance difference.
This is why I don't buy into any benchmark that is saying that the X1950XTX > the 7950GX2. I'm sure it's a great card and it's definitely priced to sell at $400 but still...
X1900XTX -> X1950XTX isn't a big jump. Faster memory, that's it. It's not enough to make a huge performance difference.
This is why I don't buy into any benchmark that is saying that the X1950XTX > the 7950GX2. I'm sure it's a great card and it's definitely priced to sell at $400 but still...
fallguy said:
My math is always right because I make sure its correct
there is a difference between being careless, and careful.Only FEAR? All of his benchmarks were low for the gx2. Its like the benchmarks were only using 1 core instead of two for most of the tests.
Daggah said:
I guess you can't understand the fact the internal 512bit memory ring on the R580 was craving more bandwidth and it got it with ddr4 memory.
Plus everyone is discounting the fact this review was using HQ which is a bad thing when benching NV cards, if you are a NV user.
Coolmanluke
Weaksauce
- Joined
- Feb 28, 2006
- Messages
- 92
Pardon me, but I thought Razor1 was banned? Every time he posts in the V-card section an argument or flame war errupts. Maybe its just me, but I thought we were better off without him.
hmm bans are temporary unless someone repeatedly provokes others in this case Fallguy started the provoking with pointing fingers in the wrong direction I just defended my position thats all, its people like you who insunate them by providing no information and just talking nonsense, well go figure.
R1ckCa1n said:
Its not a bad thing that they used HQ, but others also did test with same settings as Shimino and got double the score at least with FEAR, as showen by this http://forums.vr-zone.com/showpost.php?p=2242510&postcount=32
Yes the systems are different and all, but double the frame rates? One of the two tests just don't jive.
Yes the r580 was very bandwidth starved, but you will only see an increase when the situation is bandwidth limited as most as the increased clocks of the gddr4 which is at most ~30%. And FEAR is not a game that is heavily bandwidth limiting. This is why nV does very well in this game. The ring bus does help the r580 as its more efficient with bandwidth but still limiting factors aren't going to go away because of the increased frequencies of the vram.
razor1 said:
Perhaps you need to learn to read better. I didnt respond to you at all, you did to me.
What I did was, bring up something relevant that nobody had before, and gave it as a reason as to why the numbers may seem "off". It certainly was not in the wrong direction, and I didnt point any fingers. The fact is, HQ takes a large hit compared to Q, and that is probably a huge part of the reason that the GX2 scored lower. Could it be something else too? Yeah. But HQ has an impact. Which is why NV doesnt want reviewers using it, and suggests that they use Q. It was you who tried to "provoke" me. You had no position, as I wasnt even talking to you.
fallguy said:
Did I ever say HQ doesn't have an impact. Sorry Fallguy this is an open forum, your talk is public just as everyone else. If you want to go down this path be my guest, as for coolmannuke, both of you guys can't discuss anything without pulling out the bias card, its fun talking with both of you but unfortunately your argument is very weak.
I'm not stirring anything up. You are the ones that started to stay that I was tainted, I said, the benchmarks are being looked over by the person that did them in the first place. Thats all I said, So that puts the benchmark numbers in question. Simple.
So you were provoked by saying "any questions"? What that pissed you off because the reviewer wasn't sure if his numbers are correct. Its not your mistake you are taking it very personal.
razor1 said:
Yes its an open forum, but dont accuse me of starting anything with you, when I wasnt talking to you. Thats common sense. You also accused me of finger pointing, when again, its your who is doing that. Saying that he made a mistake with the results. I pulled out the bias card, because you are. I lean towards ATi, I can admit it. You lean towards NV, its pretty obvious. I was never "p*ssed off", dont assume you know everything. I am also not taking anything personal. Again, dont assume so much.
Perhaps the beta drivers that he used, are a problem. He claims he had the 6.7 betas, but as I recall, they werent really 6.7 betas. He used beta NV drivers as well. Thats not a good idea imo. Maybe the numbers are wrong, maybe not. My point was that HQ takes a rather large hit, over Q. And that can easily have an effect when comparing to other sites. I dont know that the scores are correct, or not. I am not taking them as factual. You think they are wrong, I dont think they are right. Im indifferent, there is a difference. At this point, they are nothing more than numbers to talk about, nothing more, nothing less.
If memory serves. the X1900XTX was supposed to have DDR4 as well, but didnt make it. Im almost postive I remember the CEO making a point about the ringbus needing faster memory, and getting a rather large boost in performance from it.
fallguy said:
Then WTF is this?
Yeah tell me what that is!?
Just because I siad the results that you are are talking about might be off you wrote that? Check Rage, I posted the link for that article over there and didn't even comment on the numbers. I thought they lined up pretty good!
Might want to check out at Rage and B3D most of my comments are very neutral and well thought out, it all depends on the situations Fallguy,
Why don't you tell me how the memory would boost performance? In what situations? Instead of talking about generalaties get specific. I know you know it, and I already said it, so go for it.
Thats my qoute at rage and this was for the ATi PR benchmarks before Vr-zone's benchmarks were released.
The numbers look way off to me, its not something like 5-10% difference is some situations thats the problem, if it was yeah the numbers could be very accurate.
PCMusicGuy
[H]ard|Gawd
- Joined
- Feb 9, 2006
- Messages
- 1,222
The benchmarks for the new ATI card look good regardless of any comparisons to other cards. If it actually comes in at the $399 dollar price point I'll certainly pick one up.
Whoa...hold on their pally. Not quite the "math" I was talking about.fallguy said:
Look at the specs. We're looking at a 550MHz boost in effective VRAM speed (a 37% clock boost). Let's assume that latency really hasn't changed much if at all. So, there we stand with a 37% clock boost out the gate. Easy math.
Now, check out the rated bandwidth. Well, we seem to be up about 29% from the XTX. Great! That's a mighty nice boost. So, what, we're going to see about a 10-15% jump in overall performance, right?
Well, that's not quite what's being reported here. A 40% boost in this title at these settings, a 37% boost in another...a 48% boost in Call of Duty 2? Hey, hold on there, bub. Something's up.
I don't care if the 7590's running at Quality, Low Quality or High Quality - the X1950's numbers are suspect as hell. Unless the latency of the GDDR4 is fantastically lower than that of the GDDR3, the numbers are way outlandishly off.
I can't take an apple and turn it into two and nor can ATi. This is the math I'm talking about, and I'm surprised you haven't bothered to question it.
EDIT: Oh, and the DailyTech reported benchmarks are looking more realistic. It's certainly showing that Quad SLi is not quite the real solution yet, and that the X1950 is fast, but not physics defying. We'll see what happens when we see some single card comparisons once the NDA's lifted - I imagine the X1950 is going to be a bitch to try and defeat.
nekrosoft13
[H]ard|Gawd
- Joined
- Jan 4, 2005
- Messages
- 1,581
skittzle said:QFT
This review shows the x1900xtx beating the 7950gx2 which goes against about everything I have read. Maybe they forgot to enable the other gpu on the 7950
excactly, that review is pure BS, 7950gx2 beats 1900XTX, it even beats 1900XT crossfire in most cases.
R1ckCa1n said:
I have my own card set to HQ...I have since I got it. I got a 2000 mark increase right off the bat in 3DMark06 from upgrading to a 7950GX2 from my X1900XT(X). Again, I don't believe that an increase in memory bandwidth is going to allow ATi to come out on top.
For what it's worth, the main reason I switched over was because of one specific issue I had with ATI's drivers: their inability to force scaling w/ aspect ratio in the drivers. I was pretty happy with my X1900XT and I don't consider myself to be particularly loyal to nVidia over ATI.
razor1 said:
Its me calling it like it is. You're NV bias, its pretty obvious from past, and present. It doesnt matter who you prefer, just dont act like you're not.
Also, vrzone used soft shadows, which take a very, very large hit. Combine soft shadows, and HQ, and frames are going to plummet for NV, and ATi.
phide said:
This isnt new news, different games, get a different performance boost from new cards.
Once again, I didnt say any of these benches are real, just that claiming they arent, based on comparing numbers to different reviews is silly. They use much different settings, that impact performance.
nekrosoft13 said:
GX2 beats a X1900 CF setup? Not hardly. Most of the time, the CF setup will be faster.
fallguy said:
Call it as you will, no more respose to your flagrent name calling.
Funny why not check that post which caused Shamino to re review his benchmarks.
Same core, different memory thats all it is, oh if you think the heat sink and fan would give a large boost think again
Silly to you, but if the settings are similiar, in a sec I'll show you a HQ test of the gx2, there would be a small varaition not a variation of close to 100%.
http://reviews.cnet.com/Nvidia_GeForce_7950_GX2/4505-8902_7-31901901-2.html
here ya go, HQ mode still doesn't lose to the x1900xt and yeah give the xt 5 fps on each benchmark and the gx 2 still won't lose.
Look at HL2 numbers the review I linked to in HQ mode the gx 2 was getting 114 fps, Shamino's benchmarks in HQ mode was getting 63 with higher resolutions. Then there are the Fear numbers which don't fit,
I didnt call you a name. Saying you're NV bias, is not calling you a name. Simple English.
You obviously think these results are not accurate. I dont think they are, and dont think they arent. Its uncorroborated numbers to me. As I said earlier, I dont read vrzone reviews. The forums are a joke, with all the anime avatars, sigs, and other nonesense. When reviews from hardware sites I like, come out, Ill take that as evidence.
I dont know what impact the much faster ram will have. Im not pretending too either. In any event, Im done talking about it with you. Feel free to carry on, I know you will.
You obviously think these results are not accurate. I dont think they are, and dont think they arent. Its uncorroborated numbers to me. As I said earlier, I dont read vrzone reviews. The forums are a joke, with all the anime avatars, sigs, and other nonesense. When reviews from hardware sites I like, come out, Ill take that as evidence.
I dont know what impact the much faster ram will have. Im not pretending too either. In any event, Im done talking about it with you. Feel free to carry on, I know you will.
InorganicMatter
[H]F Junkie
- Joined
- Oct 19, 2004
- Messages
- 15,461
I agree with Razor here, and anyone who has overclocked will as well. Memory speed is not nearly as effective on speed as core speed is. Boosting the memory by a few hundred megahertz will not boost the FPS by that same percent. Now, take that alone with the fact that VR-Zone's numbers are showing a 46% increase in FPS after a 37% increase in clock speed.
Sorry fallguy, I'm definetely not trying to advocate nVidia or put down ATI here, but I know BS when I smell it. VR-Zone published a BS "review" to get hits on their site, and it's a simple as that.
Sorry fallguy, I'm definetely not trying to advocate nVidia or put down ATI here, but I know BS when I smell it. VR-Zone published a BS "review" to get hits on their site, and it's a simple as that.
fallguy said:
So you think of thier forums like that because of avatars, just as you though of nv focus group members to be banned, wait you didn't think about that, you wanted them all to be banned, I see a pattern here. Just random name calling justifies your stance is icing on the cake.
Well you don't know what the impact of the faster ram will have, just don't talk about performance of any card then and don't try to show me that I'm wrong because I thought of it through, and don't try to call out the bias card because you didn't understand where I was coming from.
- Status