How weak is the T&L engine of the Radeon.... click to find out....

Napalm

Platinum Member
Oct 12, 1999
2,050
0
0
Synthetic benchmarks are for idiots with too much time on their hands (was that too harsh? - I can never tell ;)).

When judging a video card you should ask yourself 3 questions:

1) is it riducously fast enough for the games I play TODAY? If so, it will be quite good for the games I play tomorrow...

2) are the drivers in decent shape? If so, I can trust that it will work across platforms/games...

3) does it allow my monitor to shine? If I have a 19 inch monitor, doe I get clear text at 1600x1200 and are the colors in games nicely saturated.

Napalm
 

IBMer

Golden Member
Jul 7, 2000
1,137
0
76
Synthetic benchmarks are for idiots with too much time on their hands (was that too harsh? - I can never tell ).

Ah must be a Radeon user.

I can tell you, I own a Radeon, GTS and a V5, I am not biased. I was just incredibly suprised to find out this. When you say it will run fine in games now, but what about the high poly, T&L enabled games that will come out later?
 

lsd

Golden Member
Sep 26, 2000
1,184
70
91


<< 1) is it riducously fast enough for the games I play TODAY? If so, it will be quite good for the games I play tomorrow... >>


Says who? Yeah it *will* be able to run tommorrows games.. but at what cost? running at 640x480 with 16 bit?
 

Spoooon

Lifer
Mar 3, 2000
11,563
203
106
Anybody know of a benchmark that stresses T&amp;L that isn't considered to be pro-nVidia?
 

IBMer

Golden Member
Jul 7, 2000
1,137
0
76
I am sorry 3dMark2000 isn't Pro-NVidia. In 32-bit color the GTS and the Radeon are around the same speed + or - 500, except with the High Polygon test. If you look at ANY benchmark the GTS kills the Radeon in 16-bit color, so why should it be any different for 3DMark2000?
 

Spoooon

Lifer
Mar 3, 2000
11,563
203
106


<< 3dmark is good, but, like so many already mentioned, its (unfortunately) geared towards nvidia cards >>




<< 3dmark isnt suck... it is great for comparing differnt ram settings and such within the same hardware platform... just dont use it to gauge between different gfx cards >>



2 quotes from fellow Anandtechers. I'm not saying it is or isn't, just that that is what the general opinion is. What they base it on other than the superior results that nVidia cards get is beyond me.
 

Smbu

Platinum Member
Jul 13, 2000
2,403
0
0
It's not that 3dmark 2000 is optimized for the GTS, it's that it isn't optimized for the Radeon. It only uses 2 textures units on the cards leaving the Radeon unable to use that 3rd texture unit. BTW I have a GTS card and not a Radeon, so nobody thinks I'm biased towards the Radeon.
 

SleepyTim

Member
Oct 27, 2000
106
0
0
I always think it's interesting when someone trys to find a deep meaning in a very limited synthetic benchmark. I have a 64MB Radeon and a Creative GeForce2 Ultra in my 2 machines respectively, and I'm very familiar with the strengths and weaknesses of both cards. There are so many things wrong with this, but I just woke up so I will only name a few.

1) The synthetic benchmark in question is just that. A synthetic benchmark...... nothing more.

2) The benchmark in question was designed before the Radeon was even available, and it's not designed to measure many of the card's strengths and abilities.

3) There has been countless tests and reviews run with real games (incuding T &amp; L) that show the Radeon suffers a much less drastic performance decrease than the GTS when both the resolution and the polygon count rises. If anything, one could easily say the Radeon is better prepared for the future games than the GTS, not the other way around as a couple of you guys have suggested. I'm not necessarily saying it is, I'm only pointing out the argument is strong in the other direction too.

4) Using the principle (kinda ;) ) behind the problem with 2), we could look at the benchmarks comparing the Environment Mapped Bump Mapping (EMBM) performance of the Radeon core vs the GeForce core. Using BumpEarth: Direct3D BumpMapping, the GeForce core reaches about 5 frames per second. The Radeon core reaches a whopping 1600 frames per second running the same test. That's right.... 1600 vs 5.
In fact, the GeForce performs so poorly that even Intel's SiS630S integrated chipset beats it by a whopping 3000% with 152 frames per second in the same test. The GeForce core does not natively support EMBM, therfore it naturally gets crushed in this test. I would not start a thread about it though.

** Btw... EMBM is a very cool effect. It's a shame the GTS won't be able to keep up with the new games that will support it.

5) There are many things that make a video card a well rounded package. There is no single game, test, benchmark, (or person in a forum) that can define a card's worth.

6) Think of what would have happened if someone had started a thread called &quot;How weak is the GeForce2's 2D image quality..click to find out&quot;
Now that would have gotten ugly. (No pun intended :p)

They are both great cards though, and I am continually amazed at some people's need to prove one better in some way than the other. I take part in these discussions from time to time myself, but I pretty much spend most of my time having fun with my cards playing kick-ass games.

Anyway, just my thoughts after waking up this morning. Excuse me if they are a little.... um.... what was I saying? :D
 

PeAK

Member
Sep 25, 2000
183
0
0

The bag of &quot;tricks&quot;, sort of speak, available to
the developer of drivers to optimize framerate have
well known artifacking characteristics.
Some of you may have experienced this as stuttering in games or unacceptable framerate variation.
The link below reviews some of these questionable tatics
in the process of discussing DX8.

Warning: I am wearing my ATI hat :)
 

Dark4ng3l

Diamond Member
Sep 17, 2000
5,061
1
0
Peope made a test with the GTS and radeon a while back in Quake 3 they set the lod bias so that it loaded like 10x the normal poly load and the radeon was almost 2x as fast at 640x480 because of hsr.
 

Doomguy

Platinum Member
May 28, 2000
2,389
1
81
The Radeon's T&amp;L dosent perform nearly well as NVidia's probobly because NVidia's drivers are so mature.
 

IBMer

Golden Member
Jul 7, 2000
1,137
0
76
I got about the same responce over at the 3dfiles board. The problem with the Polygon test in 3dMark2000 uses a game engine to produce the test. Max Payne is a game coming out that uses the MaxFX engine used for 3dMark2000. The Polygon test is a low fillrate HIGH polygon test. It tests the T&amp;L engine of the graphics card.

For all of you that want to test it, try loading up Sacrifice. Turn the detail up to Insane. What you will see with the Radeon is the constant changing of the polygon models just like what I see on my V5. With the GTS the polygon models are very constant.
 

IBMer

Golden Member
Jul 7, 2000
1,137
0
76
6) Think of what would have happened if someone had started a thread called &quot;How weak is the GeForce2's 2D image quality..click to find out&quot;

It is truely sad that most of you are seeing this as a &quot;Haha the GTS is faster than the Radeon&quot; thread. If you have read any of my post in the past you would know that I stick up for the V5 a lot. This thread was truely showing that there is something wrong with the T&amp;L engine with the Radeon or something wrong in the drivers.

Now why couldn't any of you guys see it this way?
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Are some people trying to claim IBMer is a nVidiot:Q:confused:

His point is a valid one and one that has been known and shown several times on different sites, the Radeon's T&amp;L is not as strong as nVidia's. The GeForce SDR has routinely demonstrated its' edge whenever they are compared(speaking just of hardware T&amp;L).

If you don't play any T&amp;L games and don't work with any pro 3D apps then don't worry about it. If you do, you might want to know.
 

IBMer

Golden Member
Jul 7, 2000
1,137
0
76
I think weaker is an understatement. I find that the Radeon scores higher in the High Polygon tests using software T&amp;L then the Hardware T&amp;L. It is also runs better in Sacrifice with the software T&amp;L.
 

Napalm

Platinum Member
Oct 12, 1999
2,050
0
0
Ah must be a Radeon user

You may have noticed from my post (if you can read, that is) that I did not even mention the Radeon (or any other video card). I focused on your use of synthetic benchmarks that don't tell you jack-$hit about the real world.

Am I the only one who has grown weary of 14 year-old NVidiots (or adults who act like 14 year-old NVidiots) posting flame bait (i.e., like your original post) about why their video cards rock so much. Its a video card - get a grip.

Napalm
 

IBMer

Golden Member
Jul 7, 2000
1,137
0
76
You may have noticed from my post (if you can read, that is) that I did not even mention the Radeon (or any other video card). I focused on your use of synthetic benchmarks that don't tell you jack-$hit about the real world.

Am I the only one who has grown weary of 14 year-old NVidiots (or adults who act like 14 year-old NVidiots) posting flame bait (i.e., like your original post) about why their video cards rock so much. Its a video card - get a grip.


Now who is 14-years old?

Take a look at Ben's post, he laughed at the fact that someone thinks I am an NVidiot.

I was not posting this no the fact that is was more or less than any other card here. I have no clue that the T&amp;L engine on the Radeon was this weak. I felt I needed to show this to other peole who were going to buy one. I seems kind of stupid for me to just knock something I spent $250 on.

Then again, the only person in this thread that is showing any age deprivation is the only person throwing around insults of people's mentality and credibility.
 

lsd

Golden Member
Sep 26, 2000
1,184
70
91


<< Peope made a test with the GTS and radeon a while back in Quake 3 they set the lod bias so that it loaded like 10x the normal poly load and the radeon was almost 2x as fast at 640x480 because of hsr >>


The results given for that test were BS. Why? firstly they had scores for a voodoo 5 6000. Secondly, I did the test myself and I was getting ~70ish, which was 2x higher than the score given for the gf2 ultra.
 

Hawk

Platinum Member
Feb 3, 2000
2,904
0
0
Tests have been done with Quake3 with super high polygon counts, and the Radeon beat the crap out of the Geforce 2. However, the T&amp;L is not as good, that is probably due to both more immature drivers and Nvidia's &quot;experience&quot; with T&amp;L (duh, they just keep improving what they have and adding a feature each time they raise the price, I mean, release a &quot;new&quot; card).

But of course, if you run a test/benchmark that's optimized for the Radeon (or uses features only on the Radeon and not a Geforce), the Radeon will be better.

Sorry for reinterating what you guys already posted!

Point is, when &quot;tomorrow's games&quot; come out, you will have a new card and it really won't matter.

<edit>BTW, your link isn't working.</edit>
 

lsd

Golden Member
Sep 26, 2000
1,184
70
91


<< Tests have been done with Quake3 with super high polygon counts, and the Radeon beat the crap out of the Geforce 2 >>


Refer to my post above.
 

Hawk

Platinum Member
Feb 3, 2000
2,904
0
0
lsd, so therefore IBMer's comments don't make sense anymore do they? So the Geforce's T&amp;L is faster now, but what about tomorrow, with tomorrow's games? Oh, and refer to this:



<< Point is, when &quot;tomorrow's games&quot; come out, you will have a new card and it really won't matter. >>

 

sandorski

No Lifer
Oct 10, 1999
70,483
6,032
126
I too, don't put much weight to 3dmark for comparison of different vidcards. My old Viper2 beats my Voodoo5 in 3dmark easily, yet the Voodoo5 beats the Viper2 in any game out there. As for T&amp;L Nvidia vs ATI, I dunno, time will ultimately tell.

BTW, Max Payne has been cancelled.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,998
126
I click on the link and I get a:

INTERNAL ERROR
Uhoh, someone borked a cgi script. If this happens for more than 15 minutes send out an email to team@3dfiles.com. We'll take care of it pronto.


I can't comment on the issue until I see the benchmarks.
 

lsd

Golden Member
Sep 26, 2000
1,184
70
91
<< Point is, when &quot;tomorrow's games&quot; come out, you will have a new card and it really won't matter. >>
Any game that comes out now is considered &quot;tomorrow's game&quot; for my 7 month old Gladiac.