If you actually stepped foot in any tech related section of gamefaqs, you'll they are pretty much the most reputable and experienced sources you can get hardware info from. They work with software everyday.
Homebrew Why did 3dbrew...
- Thread starter MyJoyConRunsHot
- Start date
- Views 20,468
- Replies 97
- Joined
- Apr 29, 2011
- Messages
- 31,284
- Trophies
- 2
- Age
- 38
- Location
- Dr. Wahwee's castle
- XP
- 18,969
- Country
Fortunately I haven't stepped into any section of gamefaqs forums, period.
You're free to have your own opinion, I'm not saying that you're definitely wrong, I'm saying that I'd like to see some benchmarks. On paper, the 3DS uses more modern technology, but the Gamecube can perform more instructions per cycle.
I'd be grateful if you educated yourself, then by doing so. Otherwise you can go lounge in your own opinions. While I wait for facts.
You're the one making assumptions that the two CPU's are roughly the same in terms of horsepower, I'm merely questioning your opinion. Both systems run homebrew these days, all this is entirely verifiable if you want to go through the trouble of coding a benchmark and running it on both machines.
As for your original question, you asked how much of an impact will the two new cores have on the system's horsepower overall. Since they're identical to the cores the old system uses, the increase in total horsepower will be 100%. Seeing that one of the cores is almost fully used by the OS though, the perceivable improvement of performance will be higher than that in software that does use all of the cores. I hope this answers your original question.
Seems you missed the tech specs provided earlier. That's ok, it takes loads of patience and understanding to read them correctly.
Definitely software that is coded to take advantage of the additional cores will perform better, I was alluding to how exclusive games would run compared to the Wii (e.g. bigger environments, faster calculations ,etc).
A CPU designed around mobility and power efficiency does not always indicate a lack of raw power, especially when compared to far older CPU's. DMIPS has become a rather unreliable way to measure CPU performance as well. As has pure clock speed. There are CPU's even within the same type of architecture as others that thoroughly outclass those others at substantially lower clockspeeds. Like how modern AMD CPU's need multiple gigahertz higher clockspeed to even come close to matching a good Intel's performance.
On paper performance isn't always indicative of real world application. The PS2 for instance was stated by Sony to be capable of 66 million polygons per second (back when polygons were the end all be all of GPU power). That was theoretical "on paper", when in reality the real world performance was nowhere near this number when any sort of texture or lighting was added, its true performance was moderately less than 1/6th of that number at best (the GC's benchmark of 8-12 million polygons per second was far more reasonable for real world performance). Sony pulled the same crap with the PSP, stating the handheld as capable of 33 million polygons per second, it was actually closer to Dreamcast polygon levels in a real world application.
While there are a lot of poor ports on the 3DS of both Wii and PS2 games, a lot of this just comes with the territory in regards to handhelds. It's not that it's indicative of hardware power, ports in general just have less effort put into them (doubly so in regards to handhelds). Many of us who gamed in the 16 bit era and onward can tell you that the GBA suffered the same fate, it was the dumping ground for mutilated SNES and Genesis ports. Despite GBA's ridiculous advantage in processing power over either console (the performance of which was in reality more than enough to make up for the GBA's absence of a dedicated sound processor in spite of claims to the contrary). Vita is also being shat on by a dump of downgraded PS2 ports (the Jak collection probably rounding up the worst of them with a horrid choppy framerate and even button lag of all things). Again despite the Vita's obvious advantage in power over the PS2 (or any console of that generation).
But then you've got the interesting case of Monster Hunter Tri Ultimate on 3DS. MH3 was already among the best lookers on the Wii in the first place (i'm not joking when the Zelda team actually stated that they were jealous of it). Capcom had taken the time previously to port their MT Framework engine to the 3DS (which they initially used to test whether they could translate the RE5 experience into a suitable form for 3DS use) and eventually used the engine to power the MH3 port. And this according to Capcom wasn't even taking full advantage of the 3DS' power at the time. The port came out not only mostly intact (you have to look very closely, but a few very minor downgrades were made to polygon counts in environmental models), but with visual upgrades. The 3DS version runs at a much smoother framerate than the Wii version even with 3D mode on (now targeting 60fps instead of Wii's 30) and with far better lighting effects (characters/monsters now have self shadowing, and cast realistic shadows on the ground instead of the circular ones on Wii).
On paper performance isn't always indicative of real world application. The PS2 for instance was stated by Sony to be capable of 66 million polygons per second (back when polygons were the end all be all of GPU power). That was theoretical "on paper", when in reality the real world performance was nowhere near this number when any sort of texture or lighting was added, its true performance was moderately less than 1/6th of that number at best (the GC's benchmark of 8-12 million polygons per second was far more reasonable for real world performance). Sony pulled the same crap with the PSP, stating the handheld as capable of 33 million polygons per second, it was actually closer to Dreamcast polygon levels in a real world application.
While there are a lot of poor ports on the 3DS of both Wii and PS2 games, a lot of this just comes with the territory in regards to handhelds. It's not that it's indicative of hardware power, ports in general just have less effort put into them (doubly so in regards to handhelds). Many of us who gamed in the 16 bit era and onward can tell you that the GBA suffered the same fate, it was the dumping ground for mutilated SNES and Genesis ports. Despite GBA's ridiculous advantage in processing power over either console (the performance of which was in reality more than enough to make up for the GBA's absence of a dedicated sound processor in spite of claims to the contrary). Vita is also being shat on by a dump of downgraded PS2 ports (the Jak collection probably rounding up the worst of them with a horrid choppy framerate and even button lag of all things). Again despite the Vita's obvious advantage in power over the PS2 (or any console of that generation).
But then you've got the interesting case of Monster Hunter Tri Ultimate on 3DS. MH3 was already among the best lookers on the Wii in the first place (i'm not joking when the Zelda team actually stated that they were jealous of it). Capcom had taken the time previously to port their MT Framework engine to the 3DS (which they initially used to test whether they could translate the RE5 experience into a suitable form for 3DS use) and eventually used the engine to power the MH3 port. And this according to Capcom wasn't even taking full advantage of the 3DS' power at the time. The port came out not only mostly intact (you have to look very closely, but a few very minor downgrades were made to polygon counts in environmental models), but with visual upgrades. The 3DS version runs at a much smoother framerate than the Wii version even with 3D mode on (now targeting 60fps instead of Wii's 30) and with far better lighting effects (characters/monsters now have self shadowing, and cast realistic shadows on the ground instead of the circular ones on Wii).
I have read the specs of both processors, I'm familiar with them, I simply disagree with your assessment.
You're making an assessment regarding the CPU on the basis of polygon pushing power which is not handled by the CPU. We're not talking about the GPU here, we're talking about the CPU, it has nothing to do with graphics.
It depends on what you're comparing, the only element i've found easily hidden by the screen size and resolution is texture quality. And even that can be analyzed if the game in question allows you to zoom in on details with the camera (like MHTri does). A small screen or resolution doesn't hide changes in framerates for instance, and usually doesn't mask changes in lighting/shader effects (it's quite easy to tell that MHTri runs at a much higher framerate and with much improved lighting/shadow effects over the Wii original). Both polygon and texture changes can be seen quite clearly if the camera angle is zoomed in close enough to see the details.
You're missing the point i'm trying to make, I was making an assessment about theoretical "on paper" power vs real world application of hardware in general. On paper does not always translate into real world application. I used Sony's claims about the vertex performance of PS2 and PSP as a famous example of how on paper specs can be misleading. The same logic of on paper vs real world power can be applied to any hardware component though, including theoretical CPU performance (AMD for instance often misleads people about their PC CPU performance).
Determining CPU power is not as simple as taking two processors and comparing their clock speeds and DMIPS side by side.
Vertex performance statistics are always done on untextured polygons with no effects enabled, the measurement is supposed to reflect the maximum amount of triangles the graphics chip is supposed to push. If the rating reflects something else, for instance textured polygons, it's usually mentioned as a variable. Those numbers aren't misleading so much as they are misinterpreted because people think games will reach results that were reached in a laboratory in perfect conditions which is never the case.
As far as DMIPS is concerned, no, it's not an accurate benchmark, but it's the only one we have right now - it's something that can be calculated and understood, anything else is conjecture until you prove it with numbers. I've said it a hundred times in this thread - the 3DS' CPU is a much more modern design, but it's a low-power chip that does not aim at performace, it aims at power efficiency and a small heat footprint. I am certain that it would be entirely capable of matching and surpassing the Gamecube's total processing power if it ran at a higher frequency, but it doesn't because the 3DS is a mobile application of the hardware and it's not supposed to be a powerhouse.
Yes to some extent. Though the misinterpretation is largely from the lack of confirmation that these numbers aren't achievable in a real game situation. I've often seen people use the 33 or 60+ million numbers cited for the PSP and PS2 respectively in an attempt by people online to prove the systems were more powerful than they actually are. Gamecube's vertex performance on the flipside was always cited as 8-12 million, which turned out to be accurate to the real world number. I still saw people try to argue the PS2 was above GC in power using these specs, ignoring theoretical vs real world. Hell i've even seen someone try to argue the PS2 was more powerful than the Xbox with some of this sort of logic-
http://www.ign.com/boards/threads/for-the-ppl-who-think-the-ps2-is-weak.47766470/
If DMIPS were an accurate benchmark, then the 3DS would be incapable of matching the GC's CPU power even if clocked as high as Gekko. Heck it would lose eve if you overclocked it to the max of what ARM11 can reach (which is around 800mhz, and even there things get unstable). The DMIPS bench states ARM11 has only slightly above half of the performance of a comparably clocked Gekko CPU. At 268mhz it would have somewhere around less than 1/3 of Gekko's speed. You'd have to overclock the ARM11 to something around 1ghz to even match the Gekko's speed. And this isn't even including the Wii, which has a further overclocked Gekko. If the 3DS' CPU were really as weak as this, you would not be able to port a high level Wii game like Monster Hunter Tri Ultimate in any remotely comparable form to the original. Let alone with the upgrades to the framerate it received in the porting process. And that was an early 3DS game Capcom used just to chip their teeth on the 3DS hardware.
Being a CPU designed for low power consumption and heat output doesn't denote an inherent weakness compared to powerful CPU hardware from years ago. Again the GBA beat the powerful CPU's of its console predecessors SNES/Genesis, and the Vita (and most modern mobile devices) curbstomp the PS2/GC/Xbox generation of consoles (and no it wasn't just the clock speed advantages in their cases, their CPU's are simply inherently more powerful architectures even at lower clock speeds). Despite any poor ports that would provide evidence to the contrary.
As far as "laboratory" polygon count of the Gamecube is concerned, I've read it's 20 million polygons wheras the 8-12 is an in-game average, and I'm willing to accept that as an accurate number.
You're forgetting that the Gekko is single-core wheras the 3DS has two cores. It would outperform the Gekko "on paper" if it was clocked a little under Gekko's frequency. That's on paper though, real life performance is a whole different matter. You're bringing up framerate of MH Tri Ultimate when the framerate has little to do with the CPU, it's more of a GPU matter. Not all games are CPU-intensive, perhaps the bottleneck on the Wii version was not the CPU at all.
- Joined
- Nov 27, 2006
- Messages
- 6,262
- Trophies
- 1
- Location
- London
- Website
- www.youtube.com
- XP
- 651
- Country
A 20 million theoretical polygon benchmark for GC is far too low for lab performance if PS2 was over 60 million under the same circumstances. Either that or the PS2's performance is too high. Compared under the same set of conditions, the GC should be well over PS2's vertex performance.
The GPU of the 3DS on paper should be similar to the Wii's when disregarding the more advanced shader effects, comparing the more classical elements such as vertex performance. The 3DS version of Tri DID receive some additional lighting and shadow improvements over the Wii version (with only slight downgrades to polygon counts), but it also ran at a higher framerate which the GPU alone wouldn't explain. I generally targets that higher framerate even in 3D mode (which takes a toll on GPU performance).
There are other ports on the 3DS that have both visual downgrades AND a downgrade in framerate. The port of Snake Eater has lower polygon counts and draw distance in spite of the superior Pica200 GPU, yet the framerate was also much choppier (locked at 20fps max compared to PS2's 30 and dropping well into the teens and single digits frequently). DKC Returns 3D was a port of the Wii version with the geometry intact but the majority of the lighting effects removed (something kind of absurd considering that the 3DS is unquestionably better at the removed effects), it also halved the framerate from Wii and drops even below 30fps at times (switching 3D off doesn't improve performance).
The 3DS' CPU does have two cores, but most of the second core is dedicated to OS functions. And according to the developers of Pokemon X/Y, the second CPU was initially not allowed to be used by games. Earlier games such as Monster Hunter Tri likely wouldn't have had access to the second core. Heck most games probably don't even bother to use the second core at all (Smash Bros is one of the few suspected of doing so as it locks out certain OS background tasks when stressed).
The GPU of the 3DS on paper should be similar to the Wii's when disregarding the more advanced shader effects, comparing the more classical elements such as vertex performance. The 3DS version of Tri DID receive some additional lighting and shadow improvements over the Wii version (with only slight downgrades to polygon counts), but it also ran at a higher framerate which the GPU alone wouldn't explain. I generally targets that higher framerate even in 3D mode (which takes a toll on GPU performance).
There are other ports on the 3DS that have both visual downgrades AND a downgrade in framerate. The port of Snake Eater has lower polygon counts and draw distance in spite of the superior Pica200 GPU, yet the framerate was also much choppier (locked at 20fps max compared to PS2's 30 and dropping well into the teens and single digits frequently). DKC Returns 3D was a port of the Wii version with the geometry intact but the majority of the lighting effects removed (something kind of absurd considering that the 3DS is unquestionably better at the removed effects), it also halved the framerate from Wii and drops even below 30fps at times (switching 3D off doesn't improve performance).
The 3DS' CPU does have two cores, but most of the second core is dedicated to OS functions. And according to the developers of Pokemon X/Y, the second CPU was initially not allowed to be used by games. Earlier games such as Monster Hunter Tri likely wouldn't have had access to the second core. Heck most games probably don't even bother to use the second core at all (Smash Bros is one of the few suspected of doing so as it locks out certain OS background tasks when stressed).
The PS2 had two co-processors specifically for vector calculations called VU0 and VU1 (Vector Unit) and an FPU in addition to the GPU (Graphics Synthesizer), hence the insanely high polycount for its time.
Resources dedicated to the OS don't matter on paper - you're talking about real life, perceivable performance, I'm talking about laboratory performance, which are two different things altogether.
