I can see a large difference in dot 3 in every game, but not EMBM, at least for the few games I have. Battlezone II looks practically identical with/without EMBM. The only noticeable difference I see is when staring at scrap shards up close, which anyone who owns the game knows is an action practically non existant during normal gameplay.
Beyond3D's review of the VisionTek XTASY GeForce4 MX440
- Thread starter Reverend
- Start date
While EMBM in it's current form may be singing on it's last verse, the effect isn't going anywhere, in fact I expect it to be realised in more games in the future. It will be realized through pixel shaders instead, and improved by pertubing cubemaps instead of 2d textures. But it's not going to replace dot3, they will walk hand in hand. I can sign under that dot3 is more important than embm, since per-pixel lighting affects just about every polygon in the world, while embm is only going to be used for a subset, for instance of water and other reflective materials.
I agree with you. This is close to my previous post. DX8 PS 1.1 is going to completely replace EMBM for most cases.
My point is, adding EMBM alone to GF4 MX will not add much value. Well, maybe for some people it will, but definitely not for most people.
Oompa Loompa
Newcomer
On 2002-02-26 17:44, Reverend wrote:
Unacceptable.
The card is targeted at entry level users, many of whom are still using Celeron 450 class machines. You should, therefore cover all speed increments from 450 - 2.2 GHz, on both AMD, Intel, and Transmeta platforms.
Ahem...
On a more serious note, surely if you one is reviewing a video card, the goal is to make all other system components as irrelevant and non-limiting as possible? Ideally you'd want an infinitely fast CPU and no system bottlenecks, such that the properties of the video card could be examined in isolation.
Unacceptable.
The card is targeted at entry level users, many of whom are still using Celeron 450 class machines. You should, therefore cover all speed increments from 450 - 2.2 GHz, on both AMD, Intel, and Transmeta platforms.
Ahem...
On a more serious note, surely if you one is reviewing a video card, the goal is to make all other system components as irrelevant and non-limiting as possible? Ideally you'd want an infinitely fast CPU and no system bottlenecks, such that the properties of the video card could be examined in isolation.
I'm sure you know I was joking in that post of mine, Oompa Loompa (nice name!) :smile:.
As for your "serious note", the "goal" of reviewing a video card is to examine how it performs in a targetted market, first and foremost. Later on, we can examine techs in detail in a separate article.
The above holds especially true in reviewing a specific vendor's card (like VisionTek's, or LeaTek's, or eVGA's) as opposed to a reference card from a IHV.
I gather from most (including your post) that comparisons are always preferred in *any* review. Some will actually depend on such "shootout reviews" to make a purchase... others do not.
It's like a car... it's mostly a matter of budget. Give me a budget and I'll look at the competition within that budget... you can look at a car review that consists of a Ferrari, a Mercs, a BMX and a plain Ford, when only the Ford falls into your budget. Reading about the other car's performance is interesting... but that's about all it will ever be and matter to you, someone with a budget... it is the car reviewer's words about the Ford that really matters to you.
I have more that I want to say about this whole "shootout masquerading as a review" but I'm tired and would like to take a nap. I'll have more to say if responses to this post requires me to do so.
As for your "serious note", the "goal" of reviewing a video card is to examine how it performs in a targetted market, first and foremost. Later on, we can examine techs in detail in a separate article.
The above holds especially true in reviewing a specific vendor's card (like VisionTek's, or LeaTek's, or eVGA's) as opposed to a reference card from a IHV.
I gather from most (including your post) that comparisons are always preferred in *any* review. Some will actually depend on such "shootout reviews" to make a purchase... others do not.
It's like a car... it's mostly a matter of budget. Give me a budget and I'll look at the competition within that budget... you can look at a car review that consists of a Ferrari, a Mercs, a BMX and a plain Ford, when only the Ford falls into your budget. Reading about the other car's performance is interesting... but that's about all it will ever be and matter to you, someone with a budget... it is the car reviewer's words about the Ford that really matters to you.
I have more that I want to say about this whole "shootout masquerading as a review" but I'm tired and would like to take a nap. I'll have more to say if responses to this post requires me to do so.
Reverend,
A few questions:
a) Why didn't you use 3DCenter's benchmark mod for Max Payne to create your own demo? The advantage amongst others would be that the minimum framerate is kept on "0".
I tried your save game file and compared to the demos in the above mod the results shouldn't be too much apart, but wouldn't a demo provide higher accuracy especially for comparisons?
b) Although I haven't yet had the chance to see anything GF4 yet in action; any card I've seen so far running F1 2001 with FSAA on, did have noise, aliasing and weird mipmaps at best (more the games fault I guess) one thing screenshots hardly were or are able to show. Was it the same on the MX440?
c) I realise that benchmarking F1 2001 through a replay should be currently the only way to do it, but I personally would have prefered a "swingman" camera and all 21 AI cars and the player's car (=22) in the replay. Isn't the one you saved presenting more like a best case scenario with that game?
d) Since some of the guys seem to be hot on debates lately on EMBM, can someone clarify what the "EnvMap=1" variable in the player's PLR file in F1 2001 stands for? *snicker*
A few questions:
a) Why didn't you use 3DCenter's benchmark mod for Max Payne to create your own demo? The advantage amongst others would be that the minimum framerate is kept on "0".
I tried your save game file and compared to the demos in the above mod the results shouldn't be too much apart, but wouldn't a demo provide higher accuracy especially for comparisons?
b) Although I haven't yet had the chance to see anything GF4 yet in action; any card I've seen so far running F1 2001 with FSAA on, did have noise, aliasing and weird mipmaps at best (more the games fault I guess) one thing screenshots hardly were or are able to show. Was it the same on the MX440?
c) I realise that benchmarking F1 2001 through a replay should be currently the only way to do it, but I personally would have prefered a "swingman" camera and all 21 AI cars and the player's car (=22) in the replay. Isn't the one you saved presenting more like a best case scenario with that game?
d) Since some of the guys seem to be hot on debates lately on EMBM, can someone clarify what the "EnvMap=1" variable in the player's PLR file in F1 2001 stands for? *snicker*
I think that is based on a cut-scene, correct? From my experience, cutscenes do not make for good benchmarking - they're either slower than actual gameplay (usually) or faster (not usually). Can't beat actual gameplay and that is always what I try to aim for.
I ran F1 2001 with a GF3 (various types) and while "noise" and aliasing exists with a GF3, I did not notice the texture corruption I experienced with the GF4 MX with AA at high rez as I did with any GF3. I'm supposing the "noise" you mentioned is because of the game's high LOD, which exists on all cards, as do the aliasing (which is partly due to the high LOD).
Not really as that saved replay is something anyone can experience. Let's assume someone is as useless as I am at F1 2001 and usually gets left behind most of the time, which means no visible cars throughout one lap :smile:. I don't think having all 21 cars in view throughout one lap is possible... but I do get what you mean.
Personally, I cannot be absolutely sure but it could be cube maps (reasonable to expect in recent racing sims) altho I cannot find PLR file or folder in my F1 2001 install.
Well, IMO, FRAPS measurement is "real" enough. If it's your "tendency" to disagree with folks over this, I can't help that.
A cutscene usually involves various camera angles that you don't get or never happens in actual gameplay. That's what the German site's way of benchmarking Max Payne involves. There is no way F1 2001's replay is anything like a cutscene.
A cutscene usually involves various camera angles that you don't get or never happens in actual gameplay. That's what the German site's way of benchmarking Max Payne involves. There is no way F1 2001's replay is anything like a cutscene.
Rev's views in the past on Dot3 with the GTS and EMBM:
http://www.ve3d.com/3dpulpit/Reviews/nVidia_GF2GTS/2.html
Now his views are:
What happened to the balance I once knew?
<font size=-1>[ This Message was edited by: SirPauly on 2002-02-27 13:03 ]</font>
http://www.ve3d.com/3dpulpit/Reviews/nVidia_GF2GTS/2.html
Now his views are:
What happened to the balance I once knew?
<font size=-1>[ This Message was edited by: SirPauly on 2002-02-27 13:03 ]</font>
F1 2000 used EMM for the heat haze effects; I'd assume that F1 2001 would use the same engine as F1 2000 so I'd guess it is still EMBM.
Similar threads
