Is Xenos better than R520?

[bbot]

Now that the details of R520 have been revealed, the question of whether or not R520 is better than Xenos can be debated.

Let the debate begin.

 

[tEd]

I don't think we have enough details about r520 to make a real prediction how good or bad it is. Actually i think we still have not many practical details how good xenos really works.

We probably never can directly compare the 2. Different chips designed for dfferent needs.

 

[Acert93]

Now that the details of R520 have been revealed, the question of whether or not R520 is better than Xenos can be debated.

Xenos is a console specific GPU. It is missing features necessary on the PC platform, is targetted to a limited range of resolutions, and lacks video decoding/encoding and shares a memory pool with the CPU (but also has dedicated eDRAM for the backbuffer) and has a specialized interface with the CPU.

R520 is a PC specific GPU designed to accellerate DX9.c software. It is designed with numerous PC resolutions and features in mind and has hardware video decoding/encoding and has a large pool of memory dedicated completely to the GPU but has a fairly limited interface to the CPU.

We know almost nothing about either of the chips realworld performance and we have nothing to guage as a comparison between the two.

Architecturally Xenos is more advanced; beyond that we have very limited data on featureset and performance. And in many respects comparing Xenos and R520 on a PC game or console game is irrelevant because they were designed for their specific markets. There may be overlap in the markets, but as the two designs show they are quite different in many respects.

Comments on Xeno's architecture may be instructive (but already covered), especially when they are similar to R520. Really, I think the question is put forth wrongly.

I think the REAL question is: What features shared between Xenos and R520 are "wins" in game performance when comparing R520 to R420.

Comparing the architectural improvements of R520 over R420 could give us a baseline for a relative understanding of the benefits of the Xenos design.

Over the next week as we learn more about the archtecture and the scheduling and such we should get an idea of the impact this has in Xenos performance. So far it has been mainly guesses; but with R520, which shares some similar design features with Xenos, we will get our first look at HOW these design changes affect performance in general.

All generalities (no R520 vs Xenos), but they should point us in the general direction of the advantages/disadvantages of such a design.

So really, this should be, "What does R520 tell us about Xenos" not "What is better".

 

[wireframe]

LOL..."let the debate begin"...Isn't the truth closer to being that not much is known about either? You make it sound like "ok, now that we have two pieces of paper we can finally know all about them." Don't you think it would be a good idea to at least see both in action a bit before making judgments? I would even go as far as saying you'd have to program for both to get a feeling about this and not just look at what others have produced on them or what they are saying.

Oh, and I don't think you can ever really compare these two. Although they are both for the same market (video game entertainment) they will live in different systems under different conditions. For example, compare what a 'GF4' in Xbox can produce to what one in a PC can achieve or compare an even lower model Geforce to Playstation 2. On paper the Playstation 2 should lose, but I severly doubt you will play a game that looks as good as the games on PS2 with a GF2 on a PC. That may seem subjective, but if you are looking to compare pure theory (what's on paper), then there can be no debate as those are hard facts. Do they matter? Not individually. It's too complex to make a list of "wins" and pass that off as a comparison.

 

[BRiT]

Is Xenos better than R520?

Yes and No.