I don't see why? You first render with MSAA enabled, then do a resolve, then blend it with whatever else you want to display. Your 3D scene is still nicely anti-aliased.
4 years later, iPhone graphics still worse than PSP
- Thread starter edepot
- Start date
edepot, comparing GPUs based on theoretical peak performance (especially if those figures aren't even confirmed) rarely leads to meaningful conclusions.
As I mentioned, though, that doesn't prevent developers from deliberately choosing a filter mode which may appear less than ideal to others, to put it nicely.
I agree about the lack of AA, but there are no filtering hacks I'm aware of.
As I mentioned, though, that doesn't prevent developers from deliberately choosing a filter mode which may appear less than ideal to others, to put it nicely.
Sure, it's only a matter of this functionality being present in the API.
iPhone 3GS and PSP Go
Xmas, did you even look at the benchmark data? Benchmarks are practical, not theoretical, and the whole point of this thread's topic. This is the third time on this forum that I found people twisting something I wrote to fit their agenda.
I am sorry, but obviously there are a lot of people on this forum who are from various companies, and/or have special interests and political agendas of their own to make. They will use slander, outright tricks, or other methods to discredit or push their own products or ideals. You can usually find out who is trying to push what in their posts, but sometimes they have hidden agendas and are determined with resources to push those agendas. When that happens it is a fruitless effort to converse because its more about taking sides and finding fault to discredit rather than a harmonious discussion. In the world of blogs and forums I have found there are even people who are hired whose primary job is to to spread good news on some products and bad news on the competition. When that happens truthfulness is secondary to profit, and that is why half of all "popular" blogs and forums contain these people. Their mission is to help the bottom line (and in some cases a special political cause).
I have multiple times been purposely misquoted to be discredited on some subject because it did not suit their agenda (on ps3 for example), and when this happens often, obviously it degrades the quality of the forum, as then it becomes a flamefest rather than a discourse. The ultimate goal is to silence or drive people who have different ideals or loyalty to products away. So with that in mind, if you really want to know what I wrote, look in my posts directly or for this thread topic, in the links:
http://www.edepot.com/iphone.html
and
http://www.edepot.com/reviews_sony_psp.html
This is better than to trust others who misquote me purposely to propagate their ulterior motives.
As for this thread topic specifically, so what if the iPhone GPU has worse graphics than the PSP? Rather than try to cover it up, just deal with it. Make a better one next time around.
Xmas, did you even look at the benchmark data? Benchmarks are practical, not theoretical, and the whole point of this thread's topic. This is the third time on this forum that I found people twisting something I wrote to fit their agenda.
I am sorry, but obviously there are a lot of people on this forum who are from various companies, and/or have special interests and political agendas of their own to make. They will use slander, outright tricks, or other methods to discredit or push their own products or ideals. You can usually find out who is trying to push what in their posts, but sometimes they have hidden agendas and are determined with resources to push those agendas. When that happens it is a fruitless effort to converse because its more about taking sides and finding fault to discredit rather than a harmonious discussion. In the world of blogs and forums I have found there are even people who are hired whose primary job is to to spread good news on some products and bad news on the competition. When that happens truthfulness is secondary to profit, and that is why half of all "popular" blogs and forums contain these people. Their mission is to help the bottom line (and in some cases a special political cause).
I have multiple times been purposely misquoted to be discredited on some subject because it did not suit their agenda (on ps3 for example), and when this happens often, obviously it degrades the quality of the forum, as then it becomes a flamefest rather than a discourse. The ultimate goal is to silence or drive people who have different ideals or loyalty to products away. So with that in mind, if you really want to know what I wrote, look in my posts directly or for this thread topic, in the links:
http://www.edepot.com/iphone.html
and
http://www.edepot.com/reviews_sony_psp.html
This is better than to trust others who misquote me purposely to propagate their ulterior motives.
As for this thread topic specifically, so what if the iPhone GPU has worse graphics than the PSP? Rather than try to cover it up, just deal with it. Make a better one next time around.
The only benchmark data I can see in this thread is from GLBenchmark running on the iPhone, specifically results from a synthetic (i.e. quite unlike "real" workloads) feature test within that benchmark. To my knowledge this benchmark has not been ported to the PSP.
I'm not trying to twist what you write, and I'm not making a statement about the relative merits of the involved GPUs at all. The point I'm making is: either compare apples to apples (no pun intended), and compare them thoroughly, or take any conclusions with a huge amount of salt.
My company affilitation can be found in my profile.
If it wasn't already obvious from the "location"
I think that's completely unfair both to Xmas and B3D in general; as he implied without much detail, you don't seem to understand the severe limitations of that benchmark. And because you're assuming it is the one and final answer (despite the fact there's no PSP equivalent), you naturally and correctly conclude negative things wrt the iPhone 3G S GPU and PowerVR in general.
The reality is more complex. First of all, that iPhone benchmark is VSync limited for certain tests as repeated several times on this forum already. And ask any PSP developer, and you'll figure out pretty rapidly that its combination of a limited feature-set and high performance means many effects are achieved in very inefficient ways, because that's the only way to do them at all. The same thing you'd do on SGX (or any DX9-level GPU in most cases) in a few ALU instructions, you'd need another pass for on the PSP - if you can do it at all.
Your webpages on the PSP and iPhone are very nice, but FWIW I thought I'd quickly point you in the direction of two things to fix: the 3G S SoC has nothing to do with the S5PC100. This is just a naive assumption propagated by a bunch of people who don't know what's going on. It's very much a fully custom SoC, engineered partially by Apple themselves. Here's the only public hint I'm aware of: http://i.cmpnet.com/eetimes/news/09/06/image6_061909.jpg
Secondly it's interesting that you mention eDRAM, because only the original chips used eDRAM - afterwards they replaced it with in-package DRAM
I'm still amazed they got high enough performance out of the bus to do that, but hey, it seems to work pretty well: http://www.semiconductorblog.com/2008/01/31/embedded-ram/psp-graphics-processor-die-micrograph/ (there's an old article I never manage to find anymore that quotes someone from Toshiba concluding it would be more efficient to remove the eDRAM and stick to 90nm than shrink to 65nm in their case)So I really think you're very mistaken to take an aggressive stance here when you're only basing your position on simple assumptions. Don't make your conclusions seem more likely than your hypothesis; it's wrong and doesn't make you appear very likely to be right.
Top PSP games approach about three million tris/sec characteristically.
Had Sony licensed MBX Pro or an even larger, custom variant, the PSP would've been more poweful, longer running on its battery, had better image qualtiy and more functionality, and been able to release ahead of the DS.
Had Sony licensed MBX Pro or an even larger, custom variant, the PSP would've been more poweful, longer running on its battery, had better image qualtiy and more functionality, and been able to release ahead of the DS.
Blazkowicz
Legend
Anyway when was the last time you touched a game on a regular console? Cause antialiasing is either rare or completely absent there too in the wide majority of cases. I guess those aren't great 3D devices either
mobile GPU (PowerVR SGX, tegra) are advertised as capable of efficient anti-aliasing (the first series because of TBDR, the latter one because of 5x CSAA), that gives me expectations, reasonable or not. I'm a nutjob often gaming at 800x600 or 1024 with anti-aliasing on a CRT monitor mind you. I would like a home console running at 640x480 with high AA and AF.
really, an accelerated interface such as on iphone leaves me wanting something clean and low latency all around, AA is a vital part of making things looking clean. one or two more gens and mobile or home devices will offer it.
Last edited by a moderator:
I won't say your expectations aren't legitimate, but isn't it really in ISVs/developers hands what they're going to use?
XBox360 with it's Xenos+eDRAM is also capable of running 4xMSAA as a example at 640*480 (without resulting to additional tiling to fit into eDRAM). How many games have AA enabled there?
Do you have a link about this topic or is it something you know?
At what cost? Would have they been able to sell a MBX-based for the same money the PSP went for?
FWIW, I think the PSP hardware is pretty good and lots of fun playing with.
Is it similar to the PS2?
In a relative sense yes. All you have to do is compare the specifications of a PS2 with a PSP.
A Pro MBX + VGP would've still been a lot more affordable in silicon than the GPU core of PSP.
Viewing various post mortems and developer presentations on PSP games and development has given me some familiarity with the system's capabilities; my estimate of polygon performance was probably somewhat generous, actually. Just visually judging the games makes obvious the general ballpark of performance it provides.
Viewing various post mortems and developer presentations on PSP games and development has given me some familiarity with the system's capabilities; my estimate of polygon performance was probably somewhat generous, actually. Just visually judging the games makes obvious the general ballpark of performance it provides.
Are any of those post mortems publicly available? I would love to see them.
The possibility of that magnitude of improvement is there for SONY's next generation handheld, the IP just doesn't come from NVIDIA. If you're looking for Tegra IP in next generation handhelds you probably will have to concentrate on NINTENDO's corner
I've seen some stories now about Nvidia getting a 'design win' for the next Nintendo handheld using Tegra. The story wasn't clear on whether they were using the current gen Tegra or next-Gen Tegra (Cortex based). I presume it would be next Gen, as the current Tegra uses Arm11 stuff that is very long in the tooth, and even ARM is recommending Cortex5 as a replacement for ARM11.
Similar threads
