<![CDATA[Beyond3D Forum - 3D Hardware, Software & Output Devices]]> http://forum.beyond3d.com/ For discussion of current 3D graphics boards and output hardware, including reviews, drivers, control software, and video players. en Sat, 25 May 2013 13:09:00 GMT vBulletin 60 http://forum.beyond3d.com/images/styles/B3DArena/misc/rss.jpg <![CDATA[Beyond3D Forum - 3D Hardware, Software & Output Devices]]> http://forum.beyond3d.com/ Nvidia GeForce 320.18 WHQL drivers http://forum.beyond3d.com/showthread.php?t=63610&goto=newpost Thu, 23 May 2013 10:42:04 GMT Nvidia has released GeForce 320.18 WHQL drivers. http://www.nvidia.com/object/winxp-320.18-whql-driver.html http://www.nvidia.com/object/winxp64-320.18-whql-driver.html http://www.nvidia.com/object/win8-win7-winvista-32bit-320.18-whql-driver.html ... Nvidia has released GeForce 320.18 WHQL drivers.

http://www.nvidia.com/object/winxp-3...ql-driver.html

http://www.nvidia.com/object/winxp64...ql-driver.html

http://www.nvidia.com/object/win8-wi...ql-driver.html

http://www.nvidia.com/object/win8-wi...ql-driver.html

Notebook drivers

http://www.nvidia.com/object/noteboo...ql-driver.html

http://www.nvidia.com/object/noteboo...ql-driver.html ]]> DSC http://forum.beyond3d.com/showthread.php?t=63610 <![CDATA[FCAT nonsense...& just plain dumb 3d-card reviews]]> http://forum.beyond3d.com/showthread.php?t=63526&goto=newpost Tue, 14 May 2013 00:50:48 GMT My first beef here is a simple one: FCAT is apparently full of **it, if you can guess what I mean. Take a look at Hardware Canuck's "review" of the the 7990.

Follow my link above to the proper pages...so why does FCAT, a nVidia software tool--a benchmarking tool, really--report via FCAT that its products report precisely 100% the number of frames the nVidia products display? In this case, FCAT doesn't report that the nVidia product is merely "better" than the AMD product at truthfully reporting to FRAPS the number of frames displayed--FCAT claims that the nVIdia product is perfect in the numbers it reports to FRAPS--while the AMD product is all over the place--sometimes only displaying 50% of the number of frames it reports to FRAPS! Here's HC's brilliant commentary about that:

Quote:
Compare and contrast the HD 7990 to the numbers posted by the GTX 690 and you’ll find a clear win in favor of NVIDIA. This goes to prove that while the HD 7990 is a fast solution according to FRAPS, a good portion of its rendered frames never make it to the screen, thus eliminating what was a considerable advantage.
It gets better...on page two of HC's "reported versus displayed FRAPs frame rates" editorial (Two whole pages of the review are devoted to tell you why you cannot believe the AMD numbers for any FRAPs benchmark run on any AMD gpu with any AMD drivers--and that a fix is coming soon from AMD!) here's what HC says...

Quote:
FRAPS is a great too but it is obviously not able to distinguish between what’s information being passed at the software level versus what is actually being displayed on the screen. This causes the HD 7990 to go from the fastest graphics card around to a solution that routinely trails NVIDIA’s GXT 690. Hopefully AMD will have a fix for this sooner rather than later.
...The implication here is that this isn't just something affecting the 7990, but rather it is something endemic to *all* AMD gpus so that when FRAPs reports that AMD gpus are besting nVidia gpus in frame rates--you cannot believe it! It's difficult to understand why anyone would wish to publish something so derogatory to AMD without even seeking a comment from AMD. But the author of this strange review, SKYMTL (whatever that means as a name), does not convince me of anything except his complete and irredeemable anti-AMD bias.

It is almost laughable how gullible SKYMTL seems about this, as the site never even slightly questioned FCAT's very unlikely assertion that for 100% of the time in 100% of all the software tested, the nVidia product was perfect and always displayed exactly the number of frames it reported to FRAPs! (While the AMD product was consistently in 100% of the tested cases misrepresenting itself by as much as 50%!) I find that gullibility simply amazing--and I further submit that no one is that gullible unless they wish to be.

I simply do not believe what FCAT reported to HC in this case--but if as SKYMTL maintains, FCAT reported the nVIdia product perfect in the matter of displayed frames versus reported frames to FRAPs, those results are surely false and it is FCAT, rather than AMD, which cannot be believed (this should be a matter of common sense but apparently isn't, at least at Hardware Canucks.)

When history is forgotten it simply repeats. nVidia has in the past been guilty of an entire raft of phony software benchmarks and "optimized" (that's being generous) drivers. Apparently, with FCAT, the company is up to the same old tricks again. I'd prefer not to believe that. But one thing is absolutely certain: either SKYMTL is simply dead wrong and is reporting bogus FCAT numbers, or else FCAT itself is corrupt. If there's another answer here it isn't obvious.

It's a ridiculous assertion and easy to disprove--if the FCAT data was valid then what we might expect sometimes is to see FRAPs reporting certain software running at 30fps when *in reality* the AMD hardware would be running at 15fps! How could we know? Because at 15fps the AMD products would be skipping and stuttering so badly with dropped/skipped frames that even while FRAPs reports what ought to be a smooth 30fps we would find that software effectively unplayable (because it would "really" be running at 15fps.) That's just a ridiculous assertion and I know for a fact that my 5770 has never been so afflicted--and I'm running current drivers, too.

OK...Major beef #2....

I'm deciding between the 2GB 7850 and the 2GB 7790 for my next at-home card (the 1GB 5770 has managed to fulfill my needs at my monitor's 1920x1200 resolution so well over the last couple of years that I'm convinced I no longer need purchase a Flagship gpu--I'll leave that to the younger gen.)

So...I'm simply Binging/Googling reviews for these two products, and I note a disparity of 7850 2GB reviews but quite a few 2GB 7790 reviews--at first the 7850 was a 1GB card although that is no longer the case (I recall AMD saying that the 1GB 7850 had been discontinued in favor of the 2GB flavors--fine, because I want 2GB.) Here's where the reviews get really crummy, imo...

Some reviews (Hexus.net, for instance) actually ask in their review text, "Does an extra 1GB make any difference in the performance of the HD 7790?"

...but then, after asking that very question, they proceed to test the 2GB version under the identical conditions they tested the 1GB version, and concluded that not only was there no difference, but that people would be wise to steer clear of the 2GB HD 7790 (because AMD is by implication trying to sell you a load of crap that isn't a bit better than the 1GB version.) I don't want to give the impression that Hexus "acted alone" in this...;) Unfortunately, Hexus.net had plenty review company and lots of sites promised to review the 2GB card but simply repeated their 1GB review all over again with the 2GB card.

What's wrong with this picture? Longtime B3d readers (you know who you are...;)) will know what a trivial matter it is to test a 2GB frame buffer versus a 1GB frame buffer: you select resolutions, FSAA levels, and other 3d Image Quality effects *calculated* in advance to require in excess of a 1GB frame-buffer size. You do that and then you report the performance of the 2GB card versus the 1GB card--obviously(!), when you do things with the second gig of on-board ram, you are going to ***crush*** the performance of a 1GB card, which has to drop down to system ram and the *far slower* PCIex16 bus bandwidth because it doesn't have enough on-board ram to run games under the same settings. I didn't see any reviews that actually tested the second gig in the 2-gig 7790, although plenty of websites, as I say, led their readers to believe they were going to in the text of their 2GB 7790 reviews. Very sad. A couple of >1GB scenarios off the top of my head--possibly some 8xFSAA conditions, but without a doubt some Eyefinity conditions with very high resolutions and special effects ought to exceed the 1GB frame-buffer limit handily...(I'm certain people can think of plenty more scenarios in order to test > 1GB frame-buffer performance.)

Not only do these sites fail in their evaluation of the two-gig 7790, because evidently they can't think of a way to tax the frame buffer to that extent (> 1GB), but many of them also, for some strange reason, don't seem to have a problem with nVidia 2GB cards--no, it's just the AMD 2GBs that you should steer clear of...!...;) It's almost--well, ok, it is--funny...;) Except for the fact that a lot of people are going to be mislead...

If you demonstrate the incredible delta between 2gb frame-buffers and 1-gig frame-buffers when the gpu is utilizing > 1GB of the local frame buffer--but then you want to make comments to the effect that although there's a great performance difference between 2gigs and 1gig products when the first gig is exceeded, "the frame rates are still too low and for *practical reasons* you probably wouldn't want to purchase a 2GB version of the product"--fine! In that case you would have proven your statement by demonstrating performance levels of the 2GB vs. 1GB frame-buffer products when 1GB has been exceeded. I didn't read any reviews that actually took a 7790 and tested it under conditions specifically designed to tap that extra gig of on-board ram. Anybody have a link to a *real* 2GB 7790/7850 review? I'd be much obliged if you'd share it...;) Meanwhile I'll keep looking and report back here should I find something!

Honestly, the current crop of hardware reviews by various web sites is hugely sub-par by what was the status-quo in hardware reviews a decade ago. I've looked at a couple of really scary You-tube hardware reviews wherein the reviewer sounded all of 13 years old and pronounced the word infinite (as in Bioshock Infinite) En-fine'(long i)-ite(long i) three or four times so his audience would know it wasn't a mistake--he really didn't know how to pronounce the word--which meant he probably was deficient on the meaning of the word, too...;) Terrible, what passes for reviews these days. Shameful, even.

Last comment on this FCAT thing...I wish AMD would take a much more aggressive position on this issue. I understand not wanting to upset the apple cart in terms of people's partial understanding of how individual game engines can affect frame-rate and stutter to the same degree if not more than IHV drivers. The functioning of the game engine is actually crucial in these matters and often 3d-card drivers are completely incapable of patching up either a generally poor 3d game engine, or a game engine which is specifically built from the ground up to work better and faster with a particular gpu architecture than it works with another--even though everybody is writing to the same D3d/OpenGL Windows API. It looks to me that AMD might have made the mistake of thinking people commonly know more about these issues than is actually the case--and so a little bit of general-public education seems in order to correct some rather serious misapprehensions as to just how 3d games and discrete gpus work together to render a game.

Take a look at the "Dirt" games. If they are the one(s) I'm thinking of, in which AMD products nearly always consistently outperform nVidia products, this *does not mean* that nVidia drivers suck!...;) Nor does it mean that nVidia owners will have a sub-par experience when running this game--*all* it means is that when running the Dirt benchmarks the ATI products will consistently score a bit faster than the nVidia products in terms of the reported frame-rate. That's because the game engine is more closely aligned with AMD gpu architecture than with nVidia gpu architecture--and that's something that nVidia's drivers cannot do anything about. Ever. But the reverse is also true--there are some game engines more favorable to nVidia architectures than AMD architectures--and AMD will also be unable to do anything about those games, either! AMD needs to explain all of that a lot better than it has done to date. If not, the kind of FCAT nonsense I've discovered here will fill the void. AMD needs to firmly but kindly tell it like it is with relation to game engines and their effects on competitive frame-rate based benchmarks, and to the degree that FCAT spins whopping yarns like the ones pointed out here, AMD needs to specifically address them. If not they'll just fester. ]]> WaltC http://forum.beyond3d.com/showthread.php?t=63526 Nvidia GeForce 320.14 beta drivers http://forum.beyond3d.com/showthread.php?t=63523&goto=newpost Mon, 13 May 2013 14:28:03 GMT Nvidia has released GeForce 320.14 beta drivers. http://www.geforce.com/whats-new/articles/nvidia-geforce-320-14-beta-drivers-released http://www.nvidia.com/object/winxp-320.14-beta-driver.html http://www.nvidia.com/object/winxp64-320.14-beta-driver.html ... Nvidia has released GeForce 320.14 beta drivers.

http://www.geforce.com/whats-new/art...ivers-released

http://www.nvidia.com/object/winxp-3...ta-driver.html

http://www.nvidia.com/object/winxp64...ta-driver.html

http://www.nvidia.com/object/win8-wi...ta-driver.html

http://www.nvidia.com/object/win8-wi...ta-driver.html

Notebook drivers

http://www.nvidia.com/object/noteboo...ta-driver.html

http://www.nvidia.com/object/noteboo...ta-driver.html ]]> DSC http://forum.beyond3d.com/showthread.php?t=63523 Parallel mounting graphics adapters? http://forum.beyond3d.com/showthread.php?t=63451&goto=newpost Tue, 30 Apr 2013 08:41:27 GMT Bear with me here... I must be one of the least technical people who posts here. Maybe there is a perfectly good reason nobody's been doing this for so many years. I have a vague recollection of being able to buy a riser for parallel mounting vga cards back in the days of Glide3D... but maybe I'm wrong..?

To clarify, I think that it's well past time for a parallel mount alternative mATX form factor. Obviously parallel mounting expansion cards (using some sort of L-shaped bracket? to rotate them 90°) limits the form factor considerably, from the placement of the northbridge chip to the usb and audio headers etc, to the fact that I imagine only one card could fit in place at a time.

I just can't believe that after all these years of so many people having mATX LAN-party boxes with a graphics adaptor as the only add-in card, that we are still mostly unable to build cases the size of xbox360s. Am I wrong? Are there some affordable products that fill this niche? Is there some other factor I'm not consdering? Is the whole thing an idea I really should have brought up a decade ago? I think it is, but I still would find it useful. We're a PC-in-every-room family, and I can't see why we need such clumsy cases, with the only alternatives remaining shockingly expensive. ]]> poopypoo http://forum.beyond3d.com/showthread.php?t=63451