NVIDIA GeForce 7800 GTX 512MB Review [Archive]

Dave Baumann

11-Nov-2005, 15:47

<a href="http://www.beyond3d.com/content/reviews/42/"><img border="1" src="http://www.beyond3d.com/siteimages/b3dsmall.gif" align="right" width="100" height="66"></a>NVIDIA released the G70 based 7800 Series in June, some month back now, and while they are in-between their main push to 90nm based products, which looks to be in early 2006, they have been giving their line of graphics cards a bit of a polish ready for the holiday season. While the refreshes are primarily based on the GeForce 6 Series, the flagship product is also receiving an update.

While the GeForce 7800 GTX 512MB name gives away part of the basis for the product update, it certainly obscures the rest. Here we take a look at this new 7800 GTX model to see how it differs from previous GTX's and how much performance it gains over them. <a href="http://www.beyond3d.com/content/reviews/42/">Click here to read more</a>.

tEd

14-Nov-2005, 14:28

@Dave

On the max IQ benchmarks , did you use HQ AF?

Xenus

14-Nov-2005, 14:58

Typo pixel pipelines 16 on second page

tieros

14-Nov-2005, 15:02

All the chart titles on the left hand side appear to be wrong, as well as the scale.

Also, there's a typo in the conclusion: "That said though, the levels two which the core and memory "

Maintank

14-Nov-2005, 15:04

Pretty impressive card too say the least. A pretty solid 20-25% increase in performance over their previous top end card and priced against the low volume x1800XT.

They have positioned themselves really good for Christmas.

Joe DeFuria

14-Nov-2005, 15:23

Typo pixel pipelines 16 on second page

16 pixel pipelines is not a typo.

Geo

14-Nov-2005, 15:27

All the chart titles on the left hand side appear to be wrong, as well as the scale.

You get that the fill-rate graphs are actually showing something different (tho related and can be calculated from) to what the FPS charts below them represent?

pascal

14-Nov-2005, 15:59

Page 13, "Core Temperature" are in decibels.

digitalwanderer

14-Nov-2005, 16:04

Dang it, I want you to start doing some head-to-head nVidia vs ATi comparisons Dave!

(Sorry, it had to be said :oops: )

Geo

14-Nov-2005, 16:27

Nice card, NV. :smile:

Thanks as always for the reviewage, Wavey.

Anything interesting in the fillrate numbers decreasing in several instances at 16x12?

Btw, I see that ixbt/digit-life is saying that this is not a new rev of the core, just cherry-picked since June.

Xenus

14-Nov-2005, 16:46

16 pixel pipelines is not a typo.

Well it's a technicallity because there are 24 pixel shading pipes but only 16 rops which limits it to outputing 16 pixels a clock

Joe DeFuria

14-Nov-2005, 18:02

Well it's a technicallity because there are 24 pixel shading pipes but only 16 rops which limits it to outputing 16 pixels a clock

I would say calling the part a 24 pixel pipe card (as I've seen on some sites) is wrong. It's as wrong as trying to claim that the X1600 is a 12 pixel pipe card, agree?

To me, "pixel pipe" = ROP....how many pixels can be output per clock. It certainly is important to get an idea of the shading and texturing capability, but they are not the same as a pixel pipe.

I kinda wish at this time, we would adopt the ATI "nomenclature" of w-x-y-z. So the 7800 would be someting like 16-1.5-1.5-2. (I believe.)

Junkstyle

14-Nov-2005, 20:08

Why do I care about its performance compared to a 6800 Ultra? With the effort spent benching the 6800 you could have thrown in a 1800XT 512MB and given us a more meaningful comparison. How is this going to help anyone decide between the top end ATI and the top end NVidia?

Pete

14-Nov-2005, 20:29

I have two words for this beast: ye gods!

That said, those two words are derived more from reviews that stack this 7800U against the X1800XT. The analysis of this card against it's lower-clocked sibling is too clinical to inspire much enthusiasm. Pitting it against the 256MB GTX suggests mere progress. I need to see it embarass the X1800XT in the same chart to inspire the unbridled enthusiasm due it.

BTW, it really is a feat to see a 110nm part clocked within reach of a 90nm low-k part, no? Seems pretty astonishing to me.

Thowllly

14-Nov-2005, 20:36

I would say calling the part a 24 pixel pipe card (as I've seen on some sites) is wrong. It's as wrong as trying to claim that the X1600 is a 12 pixel pipe card, agree?

To me, "pixel pipe" = ROP....how many pixels can be output per clock. It certainly is important to get an idea of the shading and texturing capability, but they are not the same as a pixel pipe.

I kinda wish at this time, we would adopt the ATI "nomenclature" of w-x-y-z. So the 7800 would be someting like 16-1.5-1.5-2. (I believe.)I don't agree, to me pixel pipe=fragment pipe. In fact, not that long ago I had not even heard the term 'fragment pipe'. I think that Dave should use the word ROP to avoid confusion. In fact, Dave seems to be confused himself; in the first table on page 4 he lists "ROP's" and "Pixel Engine" as two separate entries, with no mention of fragments (OK, I just read on, according to Dave Pixel pipe=ROP and Pixel engine=fragment pipes. Not very obvious, I think). Also, he should have listed the number of fragment pipes in the specs on page 1 and/or 2.

(btw, isn't ROP's short for "ROP is"? Shouldn't it be ROPs?)

Geo

14-Nov-2005, 20:41

BTW, it really is a feat to see a 110nm part clocked within reach of a 90nm low-k part, no? Seems pretty astonishing to me.

Well, there are two ways to deal with "astonishing". One is to applaud, and NV certainly deserves some of that today --particularly since they have been pretty consistent of late in insisting that pushing clocks to the max is not the way to go for "teh win". The other is to wonder about "the other shoe" and what we aren't seeing. Maybe we got a bitty peak at that with the rumored letter from ATI to its AIB's that they can go 650 if they like and ATI will support them. It didn't sound like there were any caveats on that, which is interesting. At least it sounded to me like they were guaranteeing that every XT delivered to the AIBs would do that.

Dave Baumann

14-Nov-2005, 21:21

On the max IQ benchmarks , did you use HQ AF?
Ehhh, good point, but no.

Why do I care about its performance compared to a 6800 Ultra? With the effort spent benching the 6800 you could have thrown in a 1800XT 512MB and given us a more meaningful comparison. How is this going to help anyone decide between the top end ATI and the top end NVidia?
:roll: Take a look most of the B3D reviews....

In fact, Dave seems to be confused himself; in the first table on page 4 he lists "ROP's" and "Pixel Engine" as two separate entries, with no mention of fragments (OK, I just read on, according to Dave Pixel pipe=ROP and Pixel engine=fragment pipes. Not very obvious, I think). Also, he should have listed the number of fragment pipes in the specs on page 1 and/or 2.
The term "Pixel Engine" is what is reported by Rivatuner (I believe it is derived from NVIDIA), but that clock domain pertains to the texture units & ALU's, or in other words the fragment "engine". "Pixel Pipe" is also the tern given to the clock domain for the ROPs.

The table on page 2 is derived directly from the data in the 3D Tables, which doesn't yet have a "parallel fragments" element in there yet, although it is an addition I'll probably add at some point.

EasyRaider

14-Nov-2005, 21:33

Damn NVidia for the naming...

Now every uneducated buyer who hears the GTX 512 MB is much faster than the old GTX will assume the difference is because of the extra memory. Then they will go out and buy a 6200 512 MB...

trinibwoy

14-Nov-2005, 21:35

Jan Groothuijse

14-Nov-2005, 22:15

Wouldn't it be nice of the tables;

example:
http://www.beyond3d.com/previews/nvidia/78512/chart/image004.gif

had more than one 'explained' axis?

Neeyik

14-Nov-2005, 22:21

The x-axis is resolution - although it's not overly obvious, there are 5 data points per line, plus the table beneath the graph has 5 columns, to point this out.

Jan Groothuijse

14-Nov-2005, 22:31

The x-axis is resolution - although it's not overly obvious, there are 5 data points per line, plus the table beneath the graph has 5 columns, to point this out.
The tables dont seem to correspond with graph...

And, what exacly is the graph saying, that the 7800GTX 512MB doesn't get bottlenecked by memory...or...? Some conclusions might be nice...

Neeyik

14-Nov-2005, 22:35

Jan Groothuijse

14-Nov-2005, 22:44

Umm, the tables do correspond with the graphs. As to interpreting them, well this is one reason why I need to finish off the "how we benchmark" article that's been in the works for...umm....far too long now but seriously though, reading the text with the tables/graphs should prove sufficient conclusions.
I see now, but i really had to read twice, just to find out what the graphs where saying, allthough if i read the text a bit more thouroughly, i might just have guessed it :wink: .

Geo

14-Nov-2005, 22:50

The tables dont seem to correspond with graph...

And, what exacly is the graph saying, that the 7800GTX 512MB doesn't get bottlenecked by memory...or...? Some conclusions might be nice...

From the tables below the graphs, try multiplying the resolution (X and Y) out and then multiply by the FPS for that resolution. Then compare that to the dot on the graph. So for the one you give, 1600x1200x88.2 = 169,344,000. Only graph says it is in millions, so 169. And the last dot on the graph hits close enuf to 169 for my comfort.

It gives you an easy at-a-glance way to see where the gpu bottleneck hits, if it hits at all. On that particular graph, clearly GTX (both flavas) are ready to keep a rockin and a rollin past that point, whereas 6800 Ultra just reached its service ceiling. . .:lol:

saf1

15-Nov-2005, 03:22

Dave Baumann

15-Nov-2005, 03:36

It was a merely a comment, a factual one at that. Hell, we've said more about other (http://www.beyond3d.com/reviews/ati/9800xt_r360/index.php?p=20) refresh products.

In other words, why does it have to bring anything new to the table other than a clock cycle increase and memory?
As a site centered around the technology its always interesting to have something new to talk about.

It just seems every review here lately with Nvidia products one puts a personal bias towards it and never gives them credit for what they "did" do.
Sorry, is

That said though, the levels two which the core and memory have been increased are very high, probably passing many peoples expectations, and it appears to be a fairly impressive feat for this 110nm part."

Not doing exactly that?

Reverend

15-Nov-2005, 03:50

While I may sound like I dislike your reivew, that is not the case. It just seems every review here lately with Nvidia products one puts a personal bias towards it and never gives them credit for what they "did" do.I think Dave, like me, is more excited about reviewing products that have innovative, new 3D technology(ies). NVIDIA probably has a good lead in terms of "refresh products" (which isn't a bad thing, given your comment about shareholders' interests) but I don't think Dave is a company analyst; he reviews hardware.

saf1

15-Nov-2005, 04:10

I think Dave, like me, is more excited about reviewing products that have innovative, new 3D technology(ies). NVIDIA probably has a good lead in terms of "refresh products" (which isn't a bad thing, given your comment about shareholders' interests) but I don't think Dave is a company analyst; he reviews hardware.

If that is the case, why review the card then? Just to keep up with the other forums? That is, if there is nothing to be excited here when reviewing any card for that matter...

AlphaWolf

15-Nov-2005, 04:18

If that is the case, why review the card then? Just to keep up with the other forums? That is, if there is nothing to be excited here when reviewing any card for that matter...

Clearly he does it so people like you could read it and go on to complain about non-existant slights against an IHV.

or maybe...

He likes to inform his readers about products, because that's what he does?

Geo

15-Nov-2005, 04:20

If that is the case, why review the card then? Just to keep up with the other forums? That is, if there is nothing to be excited here when reviewing any card for that matter...

I would imagine there are several reasons, not least of which would be you can't really know there isn't anything special going on there until you see for yourself. You might also look at the saga of Sander Sassen for where both sides agree that situation started to go off the rails so far as why some cards gets reviewed, whatever the review conclusion is. . . (of course they agreed about pretty much nothing after that point). Tho having said that, I doubt it would apply in this case --the major IHV's flagship would always merit a look-see if for no other reason than to be able to talk knowledgeably about it later. . .and to have it to bench against as well.

But, really, Wavey quoted what I thot was a quite nice little earned bouquet in their direction for the card --mightn't it be just as appropriate to ask why you feel the need to focus on the relatively mild, and factual, observation about features?

Pete

15-Nov-2005, 05:41

(btw, isn't ROP's short for "ROP is"? Shouldn't it be ROPs?)The way I learned it, you could use an apostrophe to pluralize acronyms. There are probably two schools of thought for this.

Whoops, this should probably go in the 'corrections' thread (along with that "levels two which" Dave quoted ;-)).

Mariner

15-Nov-2005, 10:08

The way I learned it, you could use an apostrophe to pluralize acronyms. There are probably two schools of thought for this.

Whoops, this should probably go in the 'corrections' thread (along with that "levels two which" Dave quoted ;-)).

I think the following image states correct apostrophe usage succinctly:

http://www.angryflower.com/bobsqu.gif

No apostrophes for pluralised acronyms! :smile:

WaltC

15-Nov-2005, 12:19

Quoted from the conclusion:
"The GeForce 7800 GTX doesn't bring anything new to the table...."

My question is, was it supposed to? Or, was it Nvidia doing what the share holders expected and cherry pick the best GPU's for use at a later time to counter other manufactures?

Not that it matters, but I think paying attention to its customer base will serve the shareholders far better...;) "Cherry picking" implies very limited quantities, doesn't it?

In other words, why does it have to bring anything new to the table other than a clock cycle increase and memory? I mean, from a dollar and sense point of view, it brought enough to the table.

So you think nV's AIB partners will sell a boatload of these things at $650-$750? I presume the cost is so high due to a distinctly low number of them being available--which goes along with your "cherry picking" theory nicely.

While I may sound like I dislike your reivew, that is not the case. It just seems every review here lately with Nvidia products one puts a personal bias towards it and never gives them credit for what they "did" do.

Why should they get more credit than is due? It's an overvolted, overclocked 7x card requiring a gigantic cooler that is available in limited quantity at an astronomical price. The concept is exactly reminiscent of what they did with the original nV30 cards and coolers--if you'll recall. The goal there was to impress with "cherry picked" benchmark PR numbers that in the end did not equate with product availability (or especially stability, in the case of nV30.)

I swear - you guys have it tough. Damned if you do, damned if you don't. Anyway - good job outside my nit picking on th econclusion. My opinion is that they didn't need to bring anything to the table other than speed and bragging rights. Both of which the share holders and board of directors expect. Just my guess anyway.

Again, it's nV customers who make or break the company, and the shareholders are just along for the ride. The comment that this approach to product release is nothing new for nV is certainly accurate to a fault. But, as usual, in order to see that you have to at least be able to recall the last several years of nV product releases. IE, this product needs to be viewed in context.

Personally, I think this card is kind of cool, and it'll be interesting to see how many actually make it to market.

Jan Groothuijse

15-Nov-2005, 12:48

Not that it matters, but I think paying attention to its customer base will serve the shareholders far better...;) "Cherry picking" implies very limited quantities, doesn't it?

So you think nV's AIB partners will sell a boatload of these things at $650-$750? I presume the cost is so high due to a distinctly low number of them being available--which goes along with your "cherry picking" theory nicely.

Why should they get more credit than is due? It's an overvolted, overclocked 7x card requiring a gigantic cooler that is available in limited quantity at an astronomical price. The concept is exactly reminiscent of what they did with the original nV30 cards and coolers--if you'll recall. The goal there was to impress with "cherry picked" benchmark PR numbers that in the end did not equate with product availability (or especially stability, in the case of nV30.)
Thats a rather strange comparison, in fact, GeForce 2 GTS and Ultra, GeForce 3 Ti500, GeForce 4 Ti4600 and Ti4800, FX 5800 Ultra, FX 5900 Ultra FX 5950 Ultra and 6800 Ultra all share that same philosophy, why did you choose to compare to an two generation back product, while another great example was more recent?

And why is this wrong? If a product is sold then there are people ready to pay the money for it, and if there aren't many people willing to pay that bill, than its no big problem that it is in limited quantities.

This is normal for high-end stuff.

Oh, and the red market, well, thats just untrue.

ondaedg

15-Nov-2005, 13:47

Walt, at what point do you stop bringing up NV30? It adds nothing to the conversation. If time is irrelevant, why not talk about a Verite, Rush, or Rage MAXX as well since you're on the topic of failed consumer cards?

On a more constructive note, overall, the review is good and thorough as ever. Because this card is simply a refresh of a current product, I think pitting it against its intended competition would have proven more interesting. Dave, as you said, it did not bring any new technology to the table. Therefore, comparing it to an X1800XT would have been, in my eyes, a more suitable basis for a review. Granted, it was neat to see how in some cases it tripled the 6800U scores. If there is anything we should be talking about right now is how Nvidia managed to triple the previous generation scores in some benches (although some of it could be attributed to the video buffer increase).

It is also nice to see an "Ultra" product that wasn't just 8% faster than the standard version, but 22 to 50% faster in nearly every case. If the pricing is within 75 dollars (I can justify paying a bit more if it performs better which it obviously does) of the X1800XT and available quantity is similar, then it would be safe to say that we have a winner.

Maintank

15-Nov-2005, 18:05

Walt, at what point do you stop bringing up NV30? It adds nothing to the conversation. If time is irrelevant, why not talk about a Verite, Rush, or Rage MAXX as well since you're on the topic of failed consumer cards?

That is simple, when Nvidia screws up again and ATI actually executes so he can bring that nugget of information out and parade it around like it matters anymore.

Walt's fascination with Nvidia is interesting on the surface but at this point it seems to be more an obsession.

Why should they get more credit than is due? It's an overvolted, overclocked 7x card requiring a gigantic cooler that is available in limited quantity at an astronomical price. The concept is exactly reminiscent of what they did with the original nV30 cards and coolers--if you'll recall. The goal there was to impress with "cherry picked" benchmark PR numbers that in the end did not equate with product availability (or especially stability, in the case of nV30.)

You think the NV30 and this card have much of anything in common? lmao
Get over it already, Nvidia spanked ATI yesterday with this card. No amount of whining about the NV30 is going to change that :D

btw I have heard availability on this card is better than ATIs x1800XT that was was supposed to show up 2 weeks ago.

saf1

15-Nov-2005, 18:46

Clearly he does it so people like you could read it and go on to complain about non-existant slights against an IHV.

or maybe...

He likes to inform his readers about products, because that's what he does?

Complain? It is more of a comment, which is what this thread is for. Correct me if I'm wrong. It is a good thing I didn't say what first hit my mind, eh?

Informs the readers, is good. As I mentioned, I just found it rather odd that he worded it as he did. In my opinion it brought what it needed, for what the manufacture wanted to accomplish. And that was to beat the X1800XT, no? Hmm...or maybe we can say the X1800XT brings nothing new to the table when compared to XL?

As to how many they set aside, ala cherry picking? Who knows, I think we all know that people will be buying these. And I would wager that some will buy two for SLI mode.

Dave Baumann

15-Nov-2005, 19:57

And that was to beat the X1800XT, no?
We're not comparing it though.

saf1

16-Nov-2005, 00:21

We're not comparing it though.

Did you read what I said? I never implied that you did compare them. It was only used as a reference.

I was asking you a question - did you say the same thing when you reviewed the X1800XT and the X1800XL ala R520: Radeon X1800 page? That the XT brought nothing new to the table? I'm sure I can guess the answer, you didn't. And yes, I did read it.

We are arguing or talking about semantics I guess. To each his own.

AlphaWolf

16-Nov-2005, 00:22

Did you read what I said? I never implied that you did compare them. It was only used as a reference.

I was asking you a question - did you say the same thing when you reviewed the X1800XT and the X1800XL ala R520: Radeon X1800 page? That the XT brought nothing new to the table? I'm sure I can guess the answer, you didn't. And yes, I did read it.

We are arguing or talking about semantics I guess. To each his own.

The x1800xt did bring new things to the table so why would he say it didn't for that product?

Perhaps if he reviewed the XT 5months after reviewing the XL...

Reverend

16-Nov-2005, 00:35

If that is the case, why review the card then? Just to keep up with the other forums? That is, if there is nothing to be excited here when reviewing any card for that matter...
Because this is a hardware review site. Whether Dave likes reviewing products that have or haven't any new technologies probably isn't as important as simply having a site exist that has a (more or less) consistent output of content.

I think this thread has gone OT.

Dave Baumann

16-Nov-2005, 00:43

Did you read what I said? I never implied that you did compare them. It was only used as a reference.
Yes, and the point the flew straight over your head is that its an irrelevant reference given the context of B3D reviews since we never compare – we looking at the technologies the new processors bring and what they are capable of in relation to their processors. Your point seems to be that “it should need to bring anything new because it met its target of beating the X1800 XT”, but that’s outside of the context of the review.

I was asking you a question - did you say the same thing when you reviewed the X1800XT and the X1800XL ala R520: Radeon X1800 page?
There wasn’t a five month gap in between the two products. And you appear to ignore we’ve said pretty much the same thing for other refresh products that are of similar time spans.

We’re now over 1.5 years since 6800 was introduced, and while they did a fantastic job (http://www.beyond3d.com/previews/nvidia/nv40/index.php?p=30) with the features and performance of that, off the top of my head the only new feature since then is TrAA, and that is where the comment stems from. Given the focus of the site, yes, the new technologies are whats interesting (architecture and/or features) – that’s why we spend often 2x-3x the amount of time / pages describing and investigating them in an “Architectural review” than we do the product reviews or refreshes; this is what most of the people who visit here are coming for.

You also appear to ignore that the conclusion also states “The GeForce 7800 GTX doesn't bring anything new to the table, but it does give the 7800 line a pre-Christmas kicker, and an impressive one at that, given that this is still based on the same chip introduced five months ago and makes those high quality options more attainable in more cases.”

saf1

16-Nov-2005, 00:44

The x1800xt did bring new things to the table so why would he say it didn't for that product?

Perhaps if he reviewed the XT 5months after reviewing the XL...

For consistency sake?

Even reviewing the cards together within a weeks time, did the XT bring anything new to the table that the XL didn't already bring?

Why not forgo the what it brought or didn't bring to the table and just say how it performed? Is that really hard?

saf1

16-Nov-2005, 01:08

Yes, and the point the flew straight over your head is that its an irrelevant reference given the context of B3D reviews since we never compare – we looking at the technologies the new processors bring and what they are capable of in relation to their processors. Your point seems to be that “it should need to bring anything new because it met its target of beating the X1800 XT”, but that’s outside of the context of the review.

It didn't fly over my head - I never said you compared the two.

We both knew this GPU had nothing to do with features. It was only about speed. Which is again why I question bringing up features. In this case, I do not think it is applicable because this card is around for one, and only, one reason.

There wasn’t a five month gap in between the two products. And you appear to ignore we’ve said pretty much the same thing for other refresh products that are of similar time spans.

It was used as a point of reference. I do not care how far apart the cards are. In the context of which I used it - I simply asked a question. Did the XT bring anything new to the table that the XL already did.

We’re now over 1.5 years since 6800 was introduced, and while they did a fantastic job (http://www.beyond3d.com/previews/nvidia/nv40/index.php?p=30) with the features and performance of that, off the top of my head the only new feature since then is TrAA, and that is where the comment stems from. Given the focus of the site, yes, the new technologies are whats interesting (architecture and/or features) – that’s way we spend often 2x-3x the amount of time / pages describing and investigating them in a “Architectural review” than we do the product reviews or refreshes; this is what most of the people who visit here are coming for.

You also appear to ignore that the conclusion also states “ The GeForce 7800 GTX doesn't bring anything new to the table, but it does give the 7800 line a pre-Christmas kicker, and an impressive one at that, given that this is still based on the same chip introduced five months ago and makes those high quality options more attainable in more cases.”

I am not looking for a 7800 is x better or y better than z brand gpu. I just raised a question as to why you would say what you did. I'm outside the feature box - you are not. You've answered. We can agree to disagree - no harm, no foul.

Joe DeFuria

16-Nov-2005, 01:17

/scratches head wondering what saf1 is on about....

WaltC

16-Nov-2005, 02:13

Thats a rather strange comparison, in fact, GeForce 2 GTS and Ultra, GeForce 3 Ti500, GeForce 4 Ti4600 and Ti4800, FX 5800 Ultra, FX 5900 Ultra FX 5950 Ultra and 6800 Ultra all share that same philosophy, why did you choose to compare to an two generation back product, while another great example was more recent?

Heh...;) It seems as if when I say anything which isn't perceived as fawning and obsequious where nV is concerned that there's always someone who objects. DaveB gets called on the carpet for a perfectly reasonable remark, divorced from its context, no less, so I'm not suprised to garner a similar sort of unjustified reaction.

I picked nV30 because, unlike nV25, nV30 in the end was officially cancelled by the company and later officially repudiated (by no less than JHH himself.) Is that enough reason for you, or would you like more?...;)

And why is this wrong? If a product is sold then there are people ready to pay the money for it, and if there aren't many people willing to pay that bill, than its no big problem that it is in limited quantities.

This is normal for high-end stuff.

I don't recall having said that it was wrong. What I said was that it was typical of nV's usual competitive approach--which I think is completely supported by your own assessment of nV's philosophy as quoted above. Right?....;)

I agree that if people wish to buy into it that's their business and it certainly doesn't bother me. But I see no reason that I should necessarily have to buy into it myself, do you? I mean, isn't it sort of obvious that I don't?

Oh, and the red market, well, thats just untrue.

Sorry, but can't cipher this remark...

Reverend

16-Nov-2005, 02:24

I just raised a question as to why you would say what you did.
Because it's the goddamn truth.

If you're saying Dave should say the same things about ATI refresh products also, then all you're trying to say (rather than make all those posts above) is that you think Dave is pro-ATI and/or anti-NVIDIA. Which doesn't belong in this thread. Go start another one about this in the Site Feedback forum.

WaltC

16-Nov-2005, 02:27

Walt, at what point do you stop bringing up NV30? It adds nothing to the conversation. If time is irrelevant, why not talk about a Verite, Rush, or Rage MAXX as well since you're on the topic of failed consumer cards?

Excuse me, but it seems to me the answer to your query is exceedingly simple...;) It was my impression that the company making the product reviewed here is nVidia, and not ATi or 3dfx or anybody else. In the context of a nV product review and in the context of forum remarks surrounding only that review and nothing else, it makes little sense to start talking about everybody else's products since it isn't everybody else, or anybody else, the review examines, is it?

Furthermore, I don't buy the idea that 3dfx's mistakes or ATi's mistakes or anybody else's mistakes serve as counters to nVidia's mistakes. IE, talking about MAXX or Fury or Rush doesn't in any way, shape, or form diminish or obfuscate nV's mistakes, does it? Or, vice-versa. Besides, it wasn't "mistakes" that were being discussed in the first place--but rather it was nV's approach to competition that was being discussed. Big difference, and no need to be so touchy...;)

AlphaWolf

16-Nov-2005, 02:31

For consistency sake?

Even reviewing the cards together within a weeks time, did the XT bring anything new to the table that the XL didn't already bring?

The fact is that the x1800 boards were reviewed at the same time (at least at B3D they were).

Why not forgo the what it brought or didn't bring to the table and just say how it performed? Is that really hard?

Why not just show a graph with performance? Perhaps just show x/10 rating with no performance graphs and information. The point of reviewing the products is giving information to the reader.

The conclusion pulls the information gathered from the review together. There are no featureset improvements here, by commenting on it he makes sure that the readers know that it wasn't an aspect of the product that he overlooked.

Why do you feel that he should gloss over the lack of new features?

Pete

16-Nov-2005, 03:48

We both knew this GPU had nothing to do with features.Sadly, the review wasn't written just for you two.

Mariner, bah, the internet wins again. :D

Jan Groothuijse

16-Nov-2005, 08:20

Heh...;) It seems as if when I say anything which isn't perceived as fawning and obsequious where nV is concerned that there's always someone who objects. DaveB gets called on the carpet for a perfectly reasonable remark, divorced from its context, no less, so I'm not suprised to garner a similar sort of unjustified reaction.

I picked nV30 because, unlike nV25, nV30 in the end was officially cancelled by the company and later officially repudiated (by no less than JHH himself.) Is that enough reason for you, or would you like more?...;)

I don't recall having said that it was wrong. What I said was that it was typical of nV's usual competitive approach--which I think is completely supported by your own assessment of nV's philosophy as quoted above. Right?....;)

I agree that if people wish to buy into it that's their business and it certainly doesn't bother me. But I see no reason that I should necessarily have to buy into it myself, do you? I mean, isn't it sort of obvious that I don't?

Sorry, but can't cipher this remark...

1st. stop winking!

"impress with "cherry picked" benchmark PR numbers" was red marked in my quote, so indeed i was off by a whole character :roll:

Every product gets cancelled sometime, and the NV30 successor was launched not far after the NV30 itself, its only logical the NV30 had a relativly short life. Again, birngin up this particular chip, serves no purpose, i would have said the same if you came up with the GeForce 3 Ti500.

"It's an overvolted, overclocked 7x card requiring a gigantic cooler that is available in limited quantity at an astronomical price." But at least you didn't say it was wrong, did you? And yes as i'll say once again, that is basicly what high-end is all about, and its not just nV that does it this way, in fact, they've tried not to do it, but the GTX wasn't fast enough compared to the "overvolted overclocked unavailable ubercooled" competition (there indeed are more ways to explain this situation).

Check my post count, i couldn't have objected before, perhaps its the way you say it that makes people press that quote button.

WaltC

16-Nov-2005, 14:20

1st. stop winking!

Sure...when you stop rolling...;) BTW, what you view as a "wink" I view as a smile. If you object to the use of emoticons, take it up with the B3d staff, I guess, because the B3d forums support their use.

Every product gets cancelled sometime, and the NV30 successor was launched not far after the NV30 itself, its only logical the NV30 had a relativly short life. Again, birngin up this particular chip, serves no purpose, i would have said the same if you came up with the GeForce 3 Ti500.

Please, let's not continue revising history, OK? The nV30 reference design as originally shipped to reviewers was almost universally reviled as buggy, hot, noisy--and s-l-o-w, especially when compared to the R300 products which shipped in the prior year and weren't any of those things. Cancellation of nV30 wasn't a normal "end life" cancellation--it was cancelled prematurely after only a few thousand units were shipped. JHH publicly stated at the time of the cancellation that shipping nV30 had been a mistake, and since that time in an interview he gave not long ago right here on B3d he called nV30 a mistake yet again (quote, unquote.) Obviously, nV25 was not ever at any time considered such a mistake by nVidia or anyone else, as I recall (I owned an nV25-based card--more than one, actually, myself.) The fact of the matter is that you can't make a silk purse out of a sow's ear, and nVidia did not begin to recover from nV3x until it shipped nV40 about a year and a half later or so, nV35 and nV38 notwithstanding. When you argue that nV30 was a normal gpu cycle for nVidia, you aren't just arguing with me and with history, you're arguing with JHH himself.

It's amazing how many times this one, single fact has to be repeated--whether I state it or JHH states it makes no difference, apparently. Those who don't wish to accept it simply won't, regardless of who says it's true. The positive note here is that nVidia learned from its mistakes and moved on. On a lighter note, I never dreamed there were actual nV30 lovers out there. Must be, because every time I mention it in passing somebody's quick to tell me--and JHH--just what an unqualified success nV30 "really" was...;)

Dave Baumann

16-Nov-2005, 14:21

Dear god, Walt, how many times have we said get over it and move on, its really tiresome from you now.

WaltC

16-Nov-2005, 15:07

Dear god, Walt, how many times have we said get over it and move on, its really tiresome from you now.

's no problem, Dave...;) I'm as tired of writing it as you are of reading it, trust me. I guess it's just the historian in me when it comes to this point, but you have my word I'll never mention nV30 again---with pleasure if not gusto!...;)

saf1

16-Nov-2005, 17:10

Because it's the goddamn truth.

As it is the truth that this card was not meant to bring anything new to the table! And that is my point.

Dave Baumann

16-Nov-2005, 17:33

As it is the truth that this card was not meant to bring anything new to the table! And that is my point.
In your pre-formed opinion. But for someone reading the review who doesn't know that it wasn't supposed to add anything extra, this would be the confirmation of that!

Pete

16-Nov-2005, 21:42

You fault Dave for printing the truth?!

incurable

17-Nov-2005, 20:41

All I know after reading a number of reviews, though not this particular one yet, is that it is this card that brakes me. Or rather its MSRP.

This marks the end of my habit of buying top of the line cards, which makes me kinda sad. I would've gotten me one for Christmas, along with some fancy kit necessary because of the need to move to PCIe, but 649 Euros is to much. I don't want to pay that.

Humus

18-Nov-2005, 03:53

I would say calling the part a 24 pixel pipe card (as I've seen on some sites) is wrong. It's as wrong as trying to claim that the X1600 is a 12 pixel pipe card, agree?

To me, "pixel pipe" = ROP....how many pixels can be output per clock. It certainly is important to get an idea of the shading and texturing capability, but they are not the same as a pixel pipe.

I kinda wish at this time, we would adopt the ATI "nomenclature" of w-x-y-z. So the 7800 would be someting like 16-1.5-1.5-2. (I believe.)

Well, technically ROP deals with pixels whereas the shader deals with fragments, I think most people who reads "pixel pipe" will think of that as fragment pipes, in the same way the technically incorrect term pixel shader means fragment shader, and frankly I think more people are interested in how many fragments it can shade than how many pixels it can rasterize since that determines final performance a lot more than ROPs, though I agree it's best to declare all pipelines. Instead of specifying it as fractions, I prefer simply writing out all directly, like the X1600 as a 12-4-4 card (12 ALU, 4 TEX, 4 ROP) and X1800 as 16-16-16.

ZioniX

18-Nov-2005, 05:18

the X1600 as a 12-4-4 card (16 ALU, 4 TEX, 4 ROP)

12 only! :wink:

AlphaWolf

18-Nov-2005, 05:21

12 only! :wink:

or would it be 24? :wink:

ZioniX

18-Nov-2005, 05:56

or would it be 24? :wink:

:grin:

Here it's more like 12 ALU pixel shader engines; with an ALU pixel shader engine consisting of the full- and mini-ALU.
As it's stated in the The Radeon X1x00 Programming Guide.

BobbyRamen

18-Nov-2005, 15:08

I just wanted to say that I have read most of the 7800 GTX 512 reviews out this week. The Beyond3D review is hands down the best of them all. Clean charts with data tables to boot. I really appreciate the effort.

So sick of those busy flash charts!

I think I'll buy me two of these cards for Xmas. Question is can I put the 7800 GTX 512 Coolers on my 6800 Ultras. I use Koolance GPU blocks for my 6800 Ultras, want to sell the 6800's with the 7800 GTX coolers to bump up the selling price... if they will fit....:roll:

Sxotty

18-Nov-2005, 16:58

Maybe that will fly in Japan, but if you put some different cooler on a card it rarely nets any benifit as people worry you fubared it.

Humus

19-Nov-2005, 00:38

12 only! :wink:

Doh! Fixed.

tweek

19-Nov-2005, 16:08

So you think nV's AIB partners will sell a boatload of these things at $650-$750? I presume the cost is so high due to a distinctly low number of them being available--which goes along with your "cherry picking" theory nicely.

Why should they get more credit than is due? It's an overvolted, overclocked 7x card requiring a gigantic cooler that is available in limited quantity at an astronomical price. The concept is exactly reminiscent of what they did with the original nV30 cards and coolers--if you'll recall. The goal there was to impress with "cherry picked" benchmark PR numbers that in the end did not equate with product availability (or especially stability, in the case of nV30.)

In case you haven't noticed, the GTX is far more available than the XT at prices that are, for the most part, within a few dollars of the XT.

I'm sure you are equally critical of Ati?..frankly, except for the bit about the nv30 your above quote could just as easily be applied to the new line of their cards..to a word.