Is the MBX more powerfull than PSP?

Status
Not open for further replies.
Jov said:
PC-Engine said:
The point which you conveniently glossed over is the fact superior technology already exist for portable gaming. PSP graphics processing technology is already outdated and inferior to boot. ;)

Is this an imaginary portable with superior tech? :D

How do you justify its superior if it never materialises into a portable console :?:

Because everyone here is comparing raw specs which are known.
 
Lazy8s said:
The SH3707 SoC from Renesas implements MBX with the SuperH architecture and reportedly doubles the geometry rates of the Dreamcast, which could characteristically do about 2 million polygons per second

Only 2 Million polygons/s ? Does the performance grows and shirnks based on the outside's temperature ?

to beat out the PSP with around 4 million in-game.

This ought to be good :).

At around half the rating of the PS2, which has measured at about 7.5 million polygons per second

Keep spinning those Performance Analyzer stats some more, they never get misquoted and stretched enough.

This is awesome: from the announced T&L performance of the PSP GPU (which includes 4 cheap parallel lights IIRC) which happens to be about half the rating of VU0 + VU1 combined (taking into account GIF-to-GS bus limitations as VU0+VU1 can push more vertices than that in theory) you go on the Performance Analyzer stats, cut those in half and deduce PSP games' polygon-count ?

I expect you to rate the GCN polygonal performance right down there with the PSP, no wait... lower than that as its T&L unit on Flipper is clocked a little bit slower than PSP's GPU.

Perhaps you forgot about the VFPU on PSP's CPU and about Gekko's SIMD FPU ?

In fillrate, the SH3707's apparent use of something of an MBX Pro gives it near 750 MPixels a second to the PSP's 648.

I was not aware that the SH3707 used a 180+ MHz PowerVR MBX chip with 4 rendering pipe-lines ;).

Ar we confusing effective fill-rate with something else ?

How useful will be this "effective" fill-rate for render-to-texture and other similar effects ?

You are assuming about 4x opaque overdraw on average which is fair, but also fair is to notice how for a whole-bunch of fill-rate heavvy effects (like render-to-texture and draw-buffer operations) you will not operate at that speed and quoting a number that varies with the average on-screen overdraw is misleading at best.
 
PC-Engine said:
Panajev2001a said:
PC-Engine said:
The PSP can only do 2.6 GFLOPS and doesn't have the image quality features of an MBX Pro. The new SH-5 can do 3.2GFLOPS and paired with an MBX Pro would rip PSP a new one! :LOL:

Fine, on December the 12th will this super 457 MHz SH-5 (124 MHz higher clock-speed, but of course SCE's crappy designers managed to bloat the CPU power consumption to make up for 124 MHz of clock-speed difference :rolleyes:) + 200 MHz MBX Pro (34 MHz faster clocked than the PSP GPU and paired with a more complex VGP [DX 8.1 capabilities verus a DX7+ Hardwired T&L unit]) be on a small cool device (with a similar high-quality screen) for about $199 at a retail store ?

Yeah, I did not think so either ;).

So "it" (as in the configuration you proposed) "would" rip PSP a new one if it were out and it was available for a similar price.

Do not worry, I am sure you will spin everything into a "Sony sucks at designing anything... hurrah for any of its competitors".

I am sure you will show that such a device with much higher clock-speed all-around (124 MHz of clock-speed advantage for the SH-5 you quoted) will also have a much longer battery life at doing any kind of workload.

I wonder if you ever stop to see how you are not helping their competitors' cause by doing all this hand-waving.

Who said anything about a portable using this tech being available for purchase? :LOL:

spinning.jpg


The point which you conveniently glossed over is the fact superior technology already exist for portable gaming. PSP graphics processing technology is already outdated and inferior to boot. ;)

Is it close to market ? No ? Then it is not available and meaningless.

You are seriously stretching what superiority means.

On one side we have a $199 device with PlayStation 2 quality graphics and on the other side we have a bunch of hot air.

Yeah, superior indeed.



Oh and btw connecting clock speeds with power consumption is flawed. What do you think fab process technology is for? ;)

You must have a bug in your eye as I was connecting the two.

So far Sony+Toshiba's 90 nm process seems to be working rather well for them. Of course in your dreamy world SCE is using the worst manufacturing process to date.

Correction: SH-5 is rated at 2.8GFLOPS at 400MHz. 8)

It is rated at 3.2 GFLOPS at ~457 MHz.

7 FP ops/clock is not a difficult concept to grasp.

400 MHz --> 2.8 GFLOPS

457 MHz --> 3.2 GFLOPS



It is also not difficult to grasp the fact that the VFPU in the PSP's CPU does 8 FP ops/clock.
 
Panajev2001a!!!!!

Nooooooo!!!

Panajev2001a said:
It is also not difficult to grasp the fact that the VFPU in the PSP's CPU does 8 FP ops/clock.

But that would mean that the PSP does more FP ops/clock?!!!!1 :oops: Someone better call the ambulance for poor PC-Engine! :LOL:
 
Panajev2001a:
Only 2 Million polygons/s ? Does the performance grows and shirnks based on the outside's temperature ?
Just a realistic measure of performance -- one that typifies game conditions on the full scale. 5 million for Le Mans is a peak figure, the kind developers mostly talk about.
Keep spinning those Performance Analyzer stats some more, they never get misquoted and stretched enough.
They represent the most characteristic performance for PS2 games that's been measured. They're the only figures that make sense when the concept of general performance under general conditions is being mentioned.
Ar we confusing effective fill-rate with something else ?
I was making a comparison and so using comparative figures.
You are assuming about 4x opaque overdraw on average which is fair, but also fair is to notice how for a whole-bunch of fill-rate heavvy effects (like render-to-texture and draw-buffer operations) you will not operate at that speed and quoting a number that varies with the average on-screen overdraw is misleading at best.
I definitely agree that it could be misleading, but I was quoting the high averages from PowerVR's documentation considering that the SH3707 is already boasting enhancement beyond what's specified for the MBX line.
 
The PSP can only do 2.6 GFLOPS and doesn't have the image quality features of an MBX Pro. The new SH-5 can do 3.2GFLOPS and paired with an MBX Pro or Bitboys G40 would rip PSP a new one!

Anyone who thinks SONY has the best mobile graphics technology isn't quite in touch with reality.

even if a handheld with an SH-5 coupled with MBX Pro or G40 was superior to PSP, I don't see such a device coming out within the next 3-6 months with lots of games in the pipeline.

the only rivals to PSP will be the next GameBoy coming in 06 or 07 and a potential Xboy. until then, PSP is the highend in handhelds, regardless of new emerging mobile technologies.
 
Is it close to market ? No ? Then it is not available and meaningless.

Well if being on the market is not close to market then no. ;)

So far Sony+Toshiba's 90 nm process seems to be working rather well for them. Of course in your dreamy world SCE is using the worst manufacturing process to date.

Yeah kinda like how EE+GS@90nm isn't fully 90nm. :LOL:

It is rated at 3.2 GFLOPS at ~457 MHz.

7 FP ops/clock is not a difficult concept to grasp.

400 MHz --> 2.8 GFLOPS

457 MHz --> 3.2 GFLOPS

A 457MHz SH-5 doesn't yet exist but of course in your fantasy world it does just so you can use the flawed clock argument. :LOL:

He does not need to, he is PC-Engine... he only needs to make quick puns.

Actually the SH-5 IP cores have already been licensed by various companies. Same with MBX and Bitboys but of course this reality doesn't jive with yours since it puts PSP's graphics tech to shame.

If you hadn't noticed, this thread is about technology but of course technology is irrelevent even on the B3D messageboards because in your fantasy world tech has to be inside a SONY console to be relevent. ;)

Phil said:
Panajev2001a!!!!!

Nooooooo!!!

Panajev2001a said:
It is also not difficult to grasp the fact that the VFPU in the PSP's CPU does 8 FP ops/clock.

But that would mean that the PSP does more FP ops/clock?!!!!1 :oops: Someone better call the ambulance for poor PC-Engine! :LOL:

2.8GFLOPS > 2.6GFLOPS no matter how you spin it. :LOL: ;)
 
PC-Engine said:
2.8GFLOPS > 2.6GFLOPS no matter how you spin it. :LOL: ;)

wait!! :oops:

PC-Engine said:
The point which you conveniently glossed over is the fact superior technology already exist for portable gaming.

The point which we conventiently glossed over is the fact "superior technology" already exists - yet you fail to admit, that the PSP has a higher FP ops/clock ratio than the superior tech you mentioned? :LOL:

"8 FP ops/clock > 7 FP ops/clock no matter how you spin it. :LOL: ;) "
 
Phil said:
PC-Engine said:
2.8GFLOPS > 2.6GFLOPS no matter how you spin it. :LOL: ;)

wait!! :oops:

PC-Engine said:
The point which you conveniently glossed over is the fact superior technology already exist for portable gaming.

The point which we conventiently glossed over is the fact "superior technology" already exists - yet you fail to admit, that the PSP has a higher FP ops/clock ratio than the superior tech you mentioned? :LOL:

"8 FP ops/clock > 7 FP ops/clock no matter how you spin it. :LOL: ;) "

PSP cpu = 333MHz SH-5 = 400MHz

400MHz > 333MHz :LOL: ;)

Let's not even talk PSP efficiency. EE anyone? ;) :LOL:

Funny how the EE has all of this theoretical performance, but was held back by the MIPS core tiny caches. :LOL:

Superior indeed. Kinda like siphoning gasoline through a straw. :oops:
 
Megadrive1988:
until then, PSP is the highend in handhelds,
The various Game Boys were also the high end in handheld gaming systems when they were alone on the market, but that didn't make their level of technology automatically satisfying.
regardless of new emerging mobile technologies.
Well, this topic is about technologies and not products, hence its use of "MBX" and not a specific MBX product in the subject.
 
PSP = handheld, SH5 = cpu

handheld > cpu :LOL: ;)

Seriously, the sheer stupidy in comparing a device built around handheld restrictions with a purely fictional "device" or technology baffles me. If we're comparing technology, at least admit that the PSP CPU has a higher FP throughoutput on a clock basis than the competing technology and therefore is doing more FP operations per clock. Not as if it would change anything... a CPU alone isn't a handheld and for effectively comparing something in production TODAY with a processing unit is rather useless without taking the full device into account (including battery-, size-, heat-, functional and other restrictions found in reality).
 
Phil said:
Seriously, the sheer stupidy in comparing a device built around handheld restrictions with a purely fictional "device" or technology baffles me. If we're comparing technology, at least admit that the PSP CPU has a higher FP throughoutput on a clock basis than the competing technology and therefore is doing more FP operations per clock. Not as if it would change anything... a CPU alone isn't a handheld and for effectively comparing something in production TODAY with a processing unit is rather useless without taking the full device into account (including battery-, size-, heat-, functional and other restrictions found in reality).

First of all SH-5 is not a fictitious technology or device. Second SH-5 is designed for low power consumption less than 400 mW at 400MHz for use in low power applications . Third yes PSP cpu can do 1FP op more than SH-5 but SH-5 was designed to run at higher clock speeds so in the end 2.8GFLOPS is still higher than 2.6 anyway you look at it. You can have 50 FP ops/clock but if your chip can only reach a certain clock frequency then you have to take that into account. You also have to take into account power consumption at X nm process. And finally FLOPS is only one factor out of many. Bottom line is Hitachi is known to make very good efficient cpus. SONY is not using EE as an example assuming they designed it.
 
Panajev2001a:
...and deduce PSP games' polygon-count ?
Actually, that rating has been roughly indicated from developer findings and game appearance as well as specs and has taken into consideration factors like power usage restrictiveness, game budgets, and development curve. I understand there's a lot of possible variance there, and the margin of error can be significant when comparing figures which are so close between the two technologies. Basically then, when considering drawing speed of polygons and pixels (but not IQ or feature set), it's at least fair to say that they're comparable at somewhere near 4 million polygons and 648 MPixels per second in-game.
 
Vysez said:
Lazy8s said:
MBX seems to be more functional with better flexibility for curved surface
How come? In what MBX CUSP seems better than PSP's own solution?
I presume Lazy8 comments are based on what Sony have demonstrated (and their presentations on what you can an cannot do) and the MBX curved surface system (of which I'm quite familiar).
And, why Bézier Patches are so important? Since we all know nobody will use them in 99% of the cases, especially now that the PSP has 32MB of RAM.
If they are cheap to use, then I suspect they will be used. Just because the PC's solutions have been, well, less than stellar doesn't mean they can't be used on other systems <shrug>

Lazy8s said:
and provided over four hours of battery life at full stress in Dell Axim X50v PDAs in reviews.
The $500 Dell Axim X50v has no optical high density media storage that puts additional strain on the battery, plus its 3D performances are miles away from PSP's.
The Intel chip uses the lowest configuration of MBX - I wouldn't use that as the benchmark.
 
This thread is giving me a headache.

Note to PC and Lazy:
You say others could make a superior handheld than Sony's.

I say:
No they CAN'T. You wanna know why? Because if they COULD, they WOULD. But they DON'T, because they know it would crash, and BURN. Badly.

You wanna know why?
It's called "branding", as in, LACK OF. It doesn't say either "Playstation" or "Gameboy" on the outside of the case, hence there won't be either developer or customer support for such a thing.

Now please be quiet, like I said, you're giving me a headache with your gibbering.
 
:!: Public Service Announcement :!:

B3D Console Talk's current signal/noise ratio: 5/95, waste-of-time status: eminent
 
Guden, we're not saying someone will put these parts together to form a complete console. We're just saying the technology is available separately and are equal or superior in their respective categories with regards to technology. Of course you would need games, marketing, etc. to lauch a console assembled from these parts but nobody is debating that especially when that's common sense anyway.
 
Simon F said:
If they are cheap to use, then I suspect they will be used. Just because the PC's solutions have been, well, less than stellar doesn't mean they can't be used on other systems <shrug>
Since you're talking about that Simon, is there any paper/article/post exposing MBX's CUSP details?
Because i'm intrigued by your confidence in the fact that curved surface can be use for any elses than, at best, topology and and the like.
If MBX's implementation adressed some of the issues CSs have, it might indeed be interesting.
 
Status
Not open for further replies.
Back
Top