@MikeB
Mike, gimme a break - I've seen footage of *playing* KZ2, you'll note the lack of reticles in the KZ footage.

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

From all the videos I have seen and from looking at the pics closly the 2nd and 5th pictures are definitly in game pics. ( I can see a reticle).

All the other pics are more than likly from Cutscenes.

_________________
Live Long and keep Amigaing!

A1200, Power Tower, TF1260 128MB RAM, 68060 Rev 6, OS3.9 BB2, HD-Floppy, Mediator TX+ PCI, Voodoo 3 3000, Soundblaster 4.1, TV Card, Spider USB, 100MBit Ethernet, 16GB CF HD, 52xCDRom.

@BrianK

I think you fail to understand my point. I know nobody who bought Pac-Man for the Atari 2600 and thought it was a good port, actually everyone seems to agree the port quality was horrible.

For example the Commodore C64 version is considered to be far better:
http://www.youtube.com/watch?v=dcjWLq9uPNM

The Pac-Man brand and marketing caused the game to sell so well for the Atari 2600. It's really bad, but totally outsold the c64 version considerably. The point is just sales doesn't per se equal game quality. For example I bought Gears of War originally due to all the over-hyping by mostly US media myself, it's not a bad game, but not if I knew this game beforehand I wouldn't have bought it. (Having said that Gears of War 2 is well better IMO, apart from online which IMO is worse despite 5 vs 5 gameplay instead of 4 vs 4).

Last edited by MikeB on 01-Jan-2009 at 09:23 PM.

@wegster

Everything is done in realtime, like the opening scene where you are able to move the camera. So gameplay and cutscenes switch immediately, the game really look that good as in those screenshots.

Some scaled down beta screengrabs:




Please don't be upset, IMO it's actually something to be happy about!

@MikeB
Let's view at 720p

link
A bit weak at geometry.

link2
A bit weak in displacement mapping and texture quality...

Last edited by Hammer on 03-Jan-2009 at 12:03 PM.
Last edited by Hammer on 01-Jan-2009 at 10:39 PM.
Last edited by Hammer on 01-Jan-2009 at 10:34 PM.
Last edited by Hammer on 01-Jan-2009 at 10:34 PM.
Last edited by Hammer on 01-Jan-2009 at 10:30 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A ECS, KS 3.2, PiStorm/RPi 3A+/Emu68)

@minator

Quote:

minator wrote: @Hammer Quote: From http://online.wsj.com/article/SB123069467545545011.html There's a book on how Sony indirectly funds Microsoft's rival console. "Playing the Fool" - How Sony inadvertently helped a competitor and lost position in the videogame market. Unfortunately it appears the WSJ has blown the story out of all proportion. The PPE in Cell is similar to the cores in the 360's processor but this is based on a respin of an existing PPC research design (nothing to do with the POWER4 or the G5). It appears this was offered to MS but it was always IBM's to offer, it's not the same as offering Cell. From what I've read MS's original plans were for something more like Cell

Refer to the future MCM Xenos and Xenon fusion package for Xbox 360. Microsoft DX software stack is split between task based and data parallel based workloads.

http://au.youtube.com/watch?v=gujpzKhWRAU&feature=related
Primary goal of the XBox is to bring PC graphics to the consoles. The overall design of Xbox 360 is like the PC.

Quote:

but IBM talked them into using the PPC cores and they had them modified to add more powerful vector units.

Some Direct3D dot instructions. D3D dot generation can be done on DX10 GPUs.
Xenos GPU is missing some DX10 features, hence the patch on the CPU.

Quote:

-- It does bring up some interesting questions though. IBM will probably be only too happy to offer Cell to other companies for the next generation, could we see an XBox720 or even a Wii2 with a Cell?

Depends on the overall Direct3D Next performance. The processor solution has to be designed for pure D3D performance in mind. SPE’s static branch support is a step backwards in that regard. SPE’s ISA is not designed as D3D GPU.

Atm, it’s Microsoft that governs the future GPU design.

NVIDIA has a study while designing their DX10.05 (CUDA) GPU i.e. what they found out that more complex shader programs are more scalar in nature.

Intel’s very wide SIMD based GPU may have a problem with this i.e it seems they didn't understand the market. ATI also !@#$ with R600's VLIW based design. The improved RV770 was based around parallel scalar workloads. NV already !@#$% with NV30's VLIW based design.

Both ATI and NV already went through SIMD, MIMD, VLIW and SIMT as part of their quest to build the fastest math processor for games (er... benchmarking/point scoring).

Can SPEs compete against GPUs in workloads that matter the most in this topic?
Does STI promote abstraction layer(with JIT) so they can change or throw away the existing ISA for a faster ISA?

Quote:

Sony will have anew uber Cell and I expect MS may end up going with a cut down Laraabee, however at 32nm the current Cell might be cool enough to go into something like Wii.

The intent of XBOX product is DirectX Box. The next Xbox is most likely based on DirectX 11.X or DirectX 12.X.

Last edited by Hammer on 02-Jan-2009 at 03:00 AM.
Last edited by Hammer on 02-Jan-2009 at 02:57 AM.
Last edited by Hammer on 02-Jan-2009 at 02:24 AM.
Last edited by Hammer on 02-Jan-2009 at 02:21 AM.
Last edited by Hammer on 01-Jan-2009 at 11:21 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A ECS, KS 3.2, PiStorm/RPi 3A+/Emu68)

@MikeB

Quote:

The point is just sales doesn't per se equal game quality.

Of course not. Game quality is in the eye of the user. But that being said just because one views 1 game as more quality than another doesn't mean the sales are there. Do companies care about quality? To a point they do because some quality brings the sales andbrings return customers. However, pulling out all stops, increases costs (more programming hours) and if it won't generate more sales and instead makes a loss. Companies want to ensure a profit or at least a break even point.

This is where the Wii comes in actually. It's a bit less graphical quality. Doesn't mean the games aren't fun. But, when there are more on the streets and development costs are less than the PS3 there better be a compelling reason why you shouldn't make your game for a market of 44M people. Also, this is why we're seeing more cross platform titles between the PS3 and 360. It's difficult to make a complex game profitable on 18M people (PS3) adding in another 26M (360) now you have a market of bigger size and bigger probability of a payback. Squaresoft saw this writing so now FFXIII is coming for the 360 too in the USA and Europe.

Last edited by BrianK on 01-Jan-2009 at 11:40 PM.

@all

Playing the Fool
How Sony inadvertently helped a competitor and lost position in the videogame market.

http://online.wsj.com/article/SB123069467545545011.html?mod=rss_Books

_________________
Pegasos2 G3 running AOS 4.1 and MorphOS 2.0
Amikit user, tinkering with Icaros VM (AROS)
EFIKA owner
Amiga 1200

@fairlanefastback

Quote:

Playing the Fool How Sony inadvertently helped a competitor and lost position in the videogame market. http://online.wsj.com/article/SB123069467545545011.html?mod=rss_Books

Someone posted this earlier. I didn't get around to reading it. The WSJ seems to make this likely more sensationalized than it is. Both CPUs are PowerPC dervied, I believe from the 970? As is the Wii's processor, G3 dervied?

While "All three of the original partners had agreed that IBM would eventually sell the Cell to other clients" -- it appears the power of the SPE didn't come along with the Xenon isn't that the unit that makes the Cell the Cell?

"The deal only got worse for Sony. Both designs were delivered on time to IBM's manufacturing division, but there was a problem with the first chip run. Microsoft had had the foresight to order backup manufacturing capacity from a third party" Seems to me that the 'deal' was worse for Sony because Microsoft was smarter and thought about lining up backup manufacturing. Perhaps Sony couldn't use a 3rd party. However, the article fails to give us any detail if the deal disallowed this.

"For Sony, the Cell processor was such a debacle that two weeks after the Playstation 3 finally appeared in stores, the company fired Ken Kutaragi, the head of its gaming unit, who had championed the Cell and built the Playstation line" (MikeB's head spins around.) Seriously -- Isn't the offical line that Kutaragi stepped down? Kutaragi talked up his love of the PS and his plans for future PS. Then a bit later Kutaragi announced his stepping down? Always seemed fishy. I really wouldn't be surprised of a behind the scenes deal which allowed Kutaragi to bow out gracefully.

The article makes it seems there might be some illegal activity here. If so it would likely be on IBM's shoulders. I'd say if Sony sued them and won then we might have something of more substance.

Also it seems the architecture of the 2 machines are different. The PS3 being Cell being the focus and the RSX more of an assist to the Cell. The 360 being more Xenos focused and the Xenon assisting the GPU? In relation I think the 360 could have been what it was with a dual core x86. Likely IBM was able to make them a lower cost offer that would suit their console needs. MIPS and MFLOPS isn't everything but in 2005 I believe the Athlons were doing about twice the MIPS at lower Ghz rating and rated in higher MFLOPS than the Xenon, if memory servers.

@Hammer

Those screenshots are too wide, maybe it's better to just link them.

Note those nice screens are from a 32 multi-player map:
http://i38.tinypic.com/23iyq0k.jpg

@minator

Quote:

Unfortunately it appears the WSJ has blown the story out of all proportion. The PPE in Cell is similar to the cores in the 360's processor but this is based on a respin of an existing PPC research design (nothing to do with the POWER4 or the G5). It appears this was offered to MS but it was always IBM's to offer, it's not the same as offering Cell.

Agreed, the Cell is very different. Core at its design is IMO really the approach to have each SPE be its own system on a chip with fast dedicated system memory and how data is being moved around the Cell's EIB ring.

IMO the bulk of R&D went into the SPEs and not really into the PPE processor itself. The Cell's strength really lies in how it's all designed to work together, rather just the sum of parts.

@MikeB

Quote:

MikeB wrote: @wegster Everything is done in realtime, like the opening scene where you are able to move the camera. So gameplay and cutscenes switch immediately, the game really look that good as in those screenshots. Some scaled down beta screengrabs: Please don't be upset, IMO it's actually something to be happy about!

I'm not upset the game looks good, I'll never be upset when a PS3 game looks decent, and at least as good as a 360 game. However, they also showed KZ way back at E3 claiming it was realtime, and it was not, and I've seen a reticle in all confirmed realtime videos of the game, which I don't see here...so it remains suspect to me.

I'll be buying KZ, as long as initial reviews are in line with what I expect them to be.

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

Websites already have their review copies.

Quote:

However, they also showed KZ way back at E3 claiming it was realtime

Who are they? You have to distinguish between PR folk and the tech guys, sadly of course the PR folk get more coverage (so devs are being bothered too much, they should code instead, but sadly often have very limited technical knowledge). It's the tech guys who confirmed all the cutscenes to be realtime.

Just take the opening scene, you drop straight and seamlessly into the battle (which is a clear indication of realtime, also a dev paused the craft scene in mid air and could zoom out towards the sky and zoom in all the way towards ground level in an earlier video):
http://www.youtube.com/watch?v=3kc0j0FncCA (spoiler alert!, hit high quality button, still bad quality footage though).

@all

Link
Already providing CPUs for the 3 consoles, optimised math libraries, visual IDE/compliers and fabricating some NVIDIA GPUs, IBM wants to enter the games market.

IBM is the closest to Microsoft in terms of software centric development ecosystem provider e.g. compliers, visual IDE, optimised math libraries, developers support.

Due to their CPUs being used in all 3 consoles, IBM has development support relationship with the game developers.

Last edited by Hammer on 02-Jan-2009 at 08:30 PM.
Last edited by Hammer on 02-Jan-2009 at 08:28 PM.
Last edited by Hammer on 02-Jan-2009 at 08:15 PM.
Last edited by Hammer on 02-Jan-2009 at 08:11 PM.
Last edited by Hammer on 02-Jan-2009 at 08:00 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A ECS, KS 3.2, PiStorm/RPi 3A+/Emu68)

@BrianK

Quote:

(SNIP for space) Also it seems the architecture of the 2 machines are different. The PS3 being Cell being the focus and the RSX more of an assist to the Cell. The 360 being more Xenos focused and the Xenon assisting the GPU? In relation I think the 360 could have been what it was with a dual core x86. Likely IBM was able to make them a lower cost offer that would suit their console needs. (SNIP)

At that time, Intel offered yet another single core Celeron and IBM offered triple core PPE and willing to customise the PPE for D3D's front end.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A ECS, KS 3.2, PiStorm/RPi 3A+/Emu68)

@Hammer

IBM getting into games? I can see why they might want to develop games. However, I'd expect them NOT to get into a 4th console hardware. This would offend the 3 companies already in the market. At least 1 of them would take their console to others.

Also, yes I realize that Intel was offering a Celeron. Imagine if Microsoft had gone with AMD. They too had compelling solutions to offer in the Athlon and dual core Athlon lines.

There are rumors that Apple will now be turning to AMD for some of their solutions. We shall see.

@BrianK

Quote:

In relation I think the 360 could have been what it was with a dual core x86. Likely IBM was able to make them a lower cost offer that would suit their console needs. MIPS and MFLOPS isn't everything but in 2005 I believe the Athlons were doing about twice the MIPS at lower Ghz rating and rated in higher MFLOPS than the Xenon, if memory servers.

x86s are designed for running control code and they're very good at it, thing is what games need is more like signal processing.

Xenon (actually Waternoose but Xenon sounds better) is a PPC modified specifically for heavy duty signal processing type stuff, it can stream data out of RAM, process it and stream it directly into the GPU. It has features like cache locking and dedicated buffers for this. The Xenon vector units are full 128bit, at the time x86s only had 64 bit vector units. Xenon also has 128 vector registers, this means you can unroll a loop to hide instruction latencies. x86 does this OoO hardware, this requires a lot more complexity and power, so would restrict the frequency.

The only desktop processor even close to Xenon at the time was a G5 - it had a full 128bit vector unit, however even at peak a dual core G5 would have only 1/2 the throughput of Xenon.

So, they could have used an x86 but it would not have anything near the throughput of Xenon. they're just not designed for the same job. Even today, they may have the frequency and 128bit vector units, but without the dedicated connection and cache locking they'd not be able to keep up.

Cell on the other had was designed from the ground up to do this and use a lot less power in the process. A single Xenon core should have the same throughput as a single SPE but I expect the SPE will beat it.

_________________
Whyzzat?

@minator

Quote:

Xenon (actually Waternoose but Xenon sounds better) is a PPC modified specifically for heavy duty signal processing type stuff, it can stream data out of RAM, process it and stream it directly into the GPU. It has features like cache locking and dedicated buffers for this. The Xenon vector units are full 128bit, at the time x86s only had 64 bit vector units

Since the Xenos is missing some DX10 features (e.g. geometry shader), Xenon (this label sounds better) was modified for the Xenos (i.e. can’t wait for R600 and it was late).

A geometry shader can generate new graphics primitives, such as points, lines, and triangles, from those primitives that were sent to the beginning of the graphics pipeline.

AMD's K8 Athlon 64s has 128bit wide FADD units. What’s missing is the 128bit FMUL and 128bit FMISC. Note why Athlon 64s beats Core 1 in SSE.

Also, K8 Athlons has 3 instruction issues has to feeds the following units;
128bit SSE FADD, 64bit SSE FMUL, 64bit SSE2 FADD, 64bit SSE2 FMUL, 64bit FMISC.
Each one has access to an instruction issue pipe.

Quote:

x86 does this OoO hardware, this requires a lot more complexity and power, so would restrict the frequency

Intel’s process tech overcomes that issue. Intel Core 2 and Pentium IV clocks beyond 3Ghz with full speed L2 cache. You can have 2.8Ghz** Core 2 Duo mobile @35watts TDP.

**Automatically overclocks to 3Ghz during single core operations.

Quote:

So, they could have used an x86 but it would not have anything near the throughput of Xenon. they're just not designed for the same job. Even today, they may have the frequency and 128bit vector units, but without the dedicated connection and cache locking they'd not be able to keep up.

Intel Core 2 and AMD K10 splits their 128bit FADD and 128bit FMUL into two separate pipelines, each having it's own instruction issue pipe. X86 approaches the problem via 3 (K10) or 4 (Core 2) instruction issues. Two instruction issues cover separate FADD and FMUL units and the remaining instructions for scalar operations.

In the end, Core 2 Duo@2.2Ghz + Geforce 8600M GT GDDR3/Radeon HD 3670 GDDR3 can run any console ported games @720p without any problems.

Care to start XBOX 360/PS3 vs mainstream PC (Games for Windows) screenshot comparisons?

Last edited by Hammer on 03-Jan-2009 at 09:29 PM.
Last edited by Hammer on 03-Jan-2009 at 01:11 PM.
Last edited by Hammer on 03-Jan-2009 at 01:03 PM.
Last edited by Hammer on 03-Jan-2009 at 12:48 PM.
Last edited by Hammer on 03-Jan-2009 at 12:43 PM.
Last edited by Hammer on 03-Jan-2009 at 12:39 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A ECS, KS 3.2, PiStorm/RPi 3A+/Emu68)

@Hammer

Quote:

Intel’s process tech overcomes that issue. Intel Core 2 and Pentium IV clocks beyond 3Ghz with full speed L2 cache.

Yes, but the core 2 wasn't around then and there's 3 cores in Xenon, a high clocked triple core P4 would have burned >200W.

You seem to be forgetting that console processors have some very serious constraints, they have to be fast, run in a strict power budget and be cheap. They all have to be consistent, they *all* have to run at the same clock speed and power budget, otherwise they go in the bin.

Intel doesn't have that problem, if they make something that runs too hot or too slow they just stick a different model number and price on it. If Xenon was sold as different models the highest would go higher then 4GHz and the 3.2 GHz version would be the el-cheapo Celeron version.

It's completely irrelevant if someone else can make it 2 years later, that's 2 years too late. Intel are trying the same thing as Xenon / Cell but if you notice it's also in-order...

_________________
Whyzzat?

@minator

Quote:

It's completely irrelevant if someone else can make it 2 years later, that's 2 years too late

XBox 360 was rushed and released on 22nd of November 2005. Core 2 Duo (Conroe) was released on 27th of July 2006.

PS3 was first released in Japan on November 11, 2006. The G80 was launched on November 8, 2006. G84 was launched on April 17, 2007.

Quote:

Intel doesn't have that problem, if they make something that runs too hot or too slow they just stick a different model number and price on it. If Xenon was sold as different models the highest would go higher then 4GHz and the 3.2 GHz version would be the el-cheapo Celeron version.

To increase yields at a target performance, Xenon's L2 cache is clocked at 1.6Ghz. 3.2Ghz PPE cores look good .... on paper. Most of Xbox 360’s performance is driven by ATI’s Xenos (includes two asymmetric GPUs).

In simple terms, Intel manufactures +100 million OoO X86s per year, while Chartered manufactures high single digit million Xenons. Intel can easily hit 10 million 3Ghz OoO X86s. Intel overcomes this fabrication issue via its industrial strength.

Quote:

Intel are trying the same thing as Xenon / Cell but if you notice it's also in-order...

They already built in-order processor array with 128 programmable stream (scalar) processors equipped GMA X3100 @500Mhz (for GM965).

There are other reasons why this 128 processor array equipped GPU !@#$ badly against Geforce 9400M IGP or Mobility Radeon HD 3200 IGP.

Last edited by Hammer on 03-Jan-2009 at 09:40 PM.
Last edited by Hammer on 03-Jan-2009 at 09:39 PM.
Last edited by Hammer on 03-Jan-2009 at 09:37 PM.
Last edited by Hammer on 03-Jan-2009 at 09:26 PM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A ECS, KS 3.2, PiStorm/RPi 3A+/Emu68)