MSI X800SE

www.bit-tech.net/review/377

 

sweet review dude..

Damn the card looks like 9800allinwonder.

Anyway good job

- Sc0rian

 

Just a shame you're now copying HardOCPs review methods. The whole 'Best playable settings' is such a subjective viewpoint that it makes the whole review, well, subjective. HardOCP used to be my only stop for reviews but since they changed I've had to look elsewhere.

I agree with only using real gameplay in games for benchmarking, but that combined a good graph of the comparitive framerates, along with some analysis of what the visual quality is like to check no-one's cheating, is still the fairest way to do it in my view.

 

I've actually been doing it this way for a few months... this is a slightly shortened format that will be used for product reviews, while reference reviews will have the full low-down with image quality examinations.

http://www.bit-tech.net/review/367/
http://www.bit-tech.net/review/369/
http://www.bit-tech.net/review/371/
http://www.bit-tech.net/article/145/
http://www.bit-tech.net/article/148/

There are just a few examples... I happen to have been on a bit of a break over christmas - my first bit of time off since February, which was very welcome, so you have probably missed the fact that I have been doing that since October.

 

...and I understand if its to reduce the number of benchmarks you have to run, I know you guys dont have the budget of some sites, but just tell us that thats the reason, not some BS about the evil hardware vendors trying to trick us. Btw, I still think bit-tech is a great site.

 

Its not a case of budget.

Bigz's review are the only ones i read now, none of this 3dmark shite

 

I dont normally read the bit-tech graphics card reviews, so apologies for saying you've just started doing it this way.

 

budget? it just takes a little longer, doesnt cost horrific amounts more.

and when did he say anything bs about hardware vendors tricking us?

 

As I said, I prefer real world gaming benchmarks too, but just not in this format.

 

"The current state of the 3D world, where Independent Hardware Vendors no longer render things in a similar fashion, dictates that even an apples-to-apples comparison is no longer a true like-for-like comparison"

Excuse the hyperbole, but thats what it sounded like to me.

 

Another great review and a useful comparison, I'm looking forward to reviews of cards that up until now have been largely vapourware

One request: Could you do a couple of benchmarks with older games? It seems the comparative performance of the newer cards differs when it comes to older games and it would be a useful point of comparison for gamers who are considering an upgrade but still have a big back collection of games

 

They aren't tricking us by any means, their architectures are a long way apart which means that they often render things in a different manner, hence why I don't think apples to apples is actually apples to apples any more.

Take Splinter Cell: Pandora Tomorrow as an example... it was originally created for the Xbox and then ported over to the PC. Thus, it is based around an NVIDIA-specific code path, due to the fact that the Xbox made use of an NVIDIA graphics processor. NVIDIA's video cards make use of a shadowing technique known as buffered shadows which, IMHO, provides more realistic shadows than what ATI can deliver. ATI are forced to use a DirectX backdoor, because their hardware does not render the shadows correctly using buffered shadows - AFAIK ATI hardware may not even support that shadowing technique fully. They actually render the title using projector shadows, which are often not quite as "soft" as the buffered shadowing technique used by NVIDIA.

The image quality is similar, but the way that they've arrived at the displayed scene is completely different, hence why I've said that an apples-to-apples isn't a true apples-to-apples comparison in a lot of cases.

FarCry is another one... do I run Shader Model 2.0b on an R420-based board, along with a Shader Model 3.0 path on an NV4x based product? They're not executing the same instructions there either, despite both Shader Model 2.0b and Shader Model 3.0 having the same "basic" limit of 512 instructions with Shader Model 2.0b lacking the likes of Flow Control, changes in precision, loops and also dynamic branching.

I'll quote Tony Tamassi on this:

edit: for those who don't know who Tony Tamassi is, he's NVIDIA's Vice President of Technical Marketing - he is more than just a marketing bod, who just regurgitates what he is told to say. He spends time listening to customer feedback, helping to define product specs, and then once the products are ready for release, he acts as NVIDIA's liason and chief technical evangelist to the rest of the world.

I met him out in San Jose at NVIDIA's editor's day, and I have to say that he is one of the most knowledgible people I've spoken to - he's come across as upfront and honest about his and his competitor's products with the facts, rather than spouting out some marketing fud.

 

Since you can get hold of a 6800GT for about £40 more, and performance is on par with the 6600GT...what makes this card so special?

Apologies if there is, I'm in a rush to read the whole review so I skipped to benchmarks and specs.

 

...Its 40 quid less.

40 quid is 40 quid and if you're going to quibble over that to satisfy your NVIDIA fanboy yearnings then more fool you.

*n

 

yeah you tell him!

(ati fanboy ^^)

 

Read the final thoughts... I said that the bundle was special, the card was good, but for the price, it's very average unfortunately. As I've said, the 6600GT is a better purchase in that performance bracket, the 6800NU can be had for around £180, which will perform better than the X800 SE consistently over the board IMHO.

 

I hear what you're saying, bigz, but I just dont think that a subjective analysis is a better way to do it. Even if 'apples to apples' doesnt exist any more, i still think running each card against each other at the same resolutions and quality settings is a more relevent comparison. Say ATI cards dont do as good a job with shadows as Nvidia does. Does that mean you have to higher the resolution to compensate?

As long as the reviewer points out the differences, ideally with some screenshot comparisons, then readers can make their own decisions about whether they prefer slightly more FPS or slightly less visual quality.

 

You dont have to, but you can, that's the point. Why do these companies develop faster cards with more visually spectacular features if you dont use them at their peak playable conditions. Surely everyone wants to have it looking as good as possible, but still playable rather than rendering 1000FPS in 460x480.

 

if this helps (probably not at all), I've just ordered a 6800le for £132 and hoping to push it nearer 6800GT speeds, unlocking a few pipelines. Obviously I won't get it all that way, because I'm a n00b, but it puts the pricing into perspective

<A88>

 

Yeah, most people do (excluding hard-core deathmatchers), but the problem is that my idea of what's a playable settings is probably different from yours, which is probably different from the reviewers. At least If you provide the readers with a comparison for each card at, say, the 3 most relevent resolutions and quality settings, then they can decide for themselves.

Another thing is that some people dont like having AA on at all, while others wouldn't play a game without it. The point is, everyone has different tastes.