iGary said:Paragraphs rule.
I got lost after the first two sentences.
Lol srry bout that gonna go back and edit it for you
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
XBox 360 Specs Released [Triple Core @ 3.2Ghz each]
- Thread starter Dippo
- Start date
- Sort by reaction score
Lol srry bout that gonna go back and edit it for you
bushgreen said:
The G5 PM and the Xbox are i think quite equal in power, consoles do tend to have less OS lag and improve over the years when the software gets more in tune with the new hardware.
there are no developer kits that have 3.2ghz G5s with 3 cores and a next generation ati card. so this means all the games have been made using only maybe something like 50% of the xbox 360s power. the fastest developer kits are 2.5Ghz with ati x800 cards.
I myself hope that actually using G5s as a developing platform means that those CPUs used in Xbox are somewhat 'compliant' with the G5 and can allow the installation of OSX.
I'm really only interested in Xbox as being a cheap Mac. We'll see if it can be somehow hacked. At least this time we'll save on cables, cause 360 just comes with the USB slots as opposed to xbox's gamepad ins which you have to make/buy an adapter cable for.
I'm really only interested in Xbox as being a cheap Mac
. We'll see if it can be somehow hacked. At least this time we'll save on cables, cause 360 just comes with the USB slots as opposed to xbox's gamepad ins which you have to make/buy an adapter cable for.
Depending on the game selection and how long before the PS3, I will wait until then, but the specs are definitely impressive. Like choosing the software before choosing the platform, it makes sense to see which games you like first before deciding between Xbox or Sony PS.
Already? There arent even any games for it now. And all the talk about games being backward compatible - they say it 'may be possible' but of course smart people will understand that it's just the news crap and it won't be possible. Xbox is based on a CISC P3 and here it's a PowerPC CPU.
It might be possible. We know the PPC can emulate x86 instructions, albeit at a 10x penalty. Given three cores running 3.2 GHz each, it's possible to emulate a 733Mhz P3. The trick might be to write drivers to emulate Nvidia to ATI hardware.@23.976 said:
The biggest blunder would to make it NOT compatible with Xbox titles.
Do you really think they're gonna bother themselves writing all that emulation software when the release date is November 2005?
@23.976 said:Do you really think they're gonna bother themselves writing all that emulation software when the release date is November 2005?
they already have the technology - it's called virtual PC, which MS bought about the time they started work on the new XBox. Maybe they got it for backwards compatibility, maybe the timing is just a coincidence.
Good point. Does anyone really think Microsoft bought Virtual PC on the merits of selling the software strictly for revenue? They bought it to own the codebase for x86 to PPC emulation, and possibly to avoid long development time or lawsuits.alandail said:
This is the biggest indication yet that XBox 360 WILL run current XBox titles. What other compelling reason other than this would Microsoft acquire Virtual PC?
So the question to this answer is yes. I will eat my left nut if this isn't the case.
Lacero said:
Microsoft bought Connectix for the Virtual Server - for virtual x86 machines. Look at VMware, very good business selling virtual machines, without a bit of PPC emulation or other cross ISA emulation.
Microsoft Eyes Virtual Server Space With Connectix Acquisition
Microsoft didnt buy the Virtual PC line for the Mac product they bought it for Virtual Server
Microsoft didnt buy the Virtual PC line for the Mac product they bought it for Virtual Server
Also, don't forget that the original Windows NT (3.1) had a builtin x86 instruction set emulator for running x86 code on Alpha, PPC and MIPS. It included the "thunking" feature so that operating system calls like I/O and graphics were passed through to the native NT system and did not need emulation. (Digital did even better, by building an optimizing translator that would dynamically generate optimized Alpha code to replace the x86 code in un-modified x86 applications. http://research.compaq.com/wrl/DECarchives/DTJ/DTJP01/DTJP01PF.PDF)
It would make sense for MS to use the Virtual PC for Mac code to improve the old NT PPC emulator, but it is hard to stretch this to be the "smoking gun" that the Connectix aquisition proves the BC question.
The "thunking" technology, however, is quite relevant to the question of emulation. When you realize that a lot of the work of the game is done in the system APIs (like DirectX) which can be native code, the prospect of emulating the x86 code of the game itself is less daunting. Much of what the 733 MHz x86 was doing can be done in the native PPC system calls.
but... doesn't anyone think it lags like heck, VirtualPC? I mean, running a game on it... Gimme a break. They'll have to do better than that.
G.Kirby said:I hope this makes it easier to port more games to Mac. If this is the case this will be great for Apple. Bring on the heavy hitting titles, Half Life 3, UT 2006 etc.
Then why didn't the GameCube (with it's proc based on the G3) bring more games to the mac then? The problem with PC games is not the x86 to PPC challenge, but the DirectX to OpenGL conversion.
Most of your post was doing quite well but most of this, which should have been a separate paragraph by the way, is completely wrong.SurfAddict said:
1 core on one die = 1 processor. 3 cores on a die = 1 physical processor. One piece of silicon = 1 physical processor, it might just be a multiple core processor. Now it can handle multiple threads, and in fact this is Microsoft's plan, offload AI to one, physics to one and whatever to the last.
Microsoft shows separate hardware based threads in their OS as separate processors (for reference they are called logical processors) but it isn't actually multiple physical processors.
Multithreading is essentially the duplication of some bits of the processor in order to ensure a better utilisation of a core or processor as a whole. Again it allows separate threads to run on a single processor but it doesn't mean it is actually a new processor is is just an additional process occurring simultaneously. Multithreading actually shares some resources but it can allow two threads to run simultaneously. It runs at full clock speed still.
Edit: Rather than just rewrite what somebody has already said you can head to Ars to read up on multithreading and multitasking in general.
As for the whole Multi threaded / chipped / etc debate.
One DIE on one chip = 1 processor
One DIE on one chip W/ Hyperthreading = 2 logical processors (with logical being somewhat less than having another "real" processor)
2 or more DIES on ONE CHIP = 2 or more REAL processors (think of it as having 2 processors, just not on seperate chips... each die is a full processor, with its own cache, etc.)
So yes, this is basically the same as a 3 processor system. There are less interconnects and seperate components by putting it on one chip.
Thats all.
One DIE on one chip = 1 processor
One DIE on one chip W/ Hyperthreading = 2 logical processors (with logical being somewhat less than having another "real" processor)
2 or more DIES on ONE CHIP = 2 or more REAL processors (think of it as having 2 processors, just not on seperate chips... each die is a full processor, with its own cache, etc.)
So yes, this is basically the same as a 3 processor system. There are less interconnects and seperate components by putting it on one chip.
Thats all.
wrong, wrong, wrong - just plain wrong
This is completely wrong.
A "core" is a complete processor, with registers, (usually L1, sometimes L1+L2) cache, execution units, and everything that one normally associates with a "processor".
SMT (or HyperThreading) is lighter weight - it has a unique set of registers and processor state for each thread, but shared execution units and cache.
The Xbox 360 chip has three cores and 6 threads - there are three complete sets of execution units, three L1 caches, six sets of registers, and six sets of processor states.
It *is* three hyperthreaded processors. The fact that all three processors (and all six threads) have been put on one piece of silicon is mostly irrelevant. ("Mostly" because the L2 cache is directly shared on-chip.)
Please, spend some time to understand the differences between SMT (hyperthreading), SMP and multiple cores before posting.
Telomar said:
This is completely wrong.
A "core" is a complete processor, with registers, (usually L1, sometimes L1+L2) cache, execution units, and everything that one normally associates with a "processor".
SMT (or HyperThreading) is lighter weight - it has a unique set of registers and processor state for each thread, but shared execution units and cache.
The Xbox 360 chip has three cores and 6 threads - there are three complete sets of execution units, three L1 caches, six sets of registers, and six sets of processor states.
It *is* three hyperthreaded processors. The fact that all three processors (and all six threads) have been put on one piece of silicon is mostly irrelevant. ("Mostly" because the L2 cache is directly shared on-chip.)
Please, spend some time to understand the differences between SMT (hyperthreading), SMP and multiple cores before posting.
3 hyperthreaded cores is just that, 3 hyperthreaded cores in 1 processor. They still only form a single discrete processor though, albeit now one that has support for multiple threads in hardware. At no stage did I state that SMT and multiple cores were not different, in fact I specifically stated that SMT shares resources, part of which includes some of the registers. You don't add a completely duplicated set of all the registers.AidenShaw said:
A processor continues to be a single integrated part of silicon, a core is only a part of that silicon.
Sorry if you can't understand the distinction but the distinction is there and becomes very important on certain pieces of software where licenses are by processor or physical thread. The difference = very big money.
Edit: Exhibit 1. Microsoft themselves consider it a single processor. You can argue with that until you turn blue in the face but you'll still be wrong *shrug*
Telomar said:
The only part I was really wrong about is that I meant 3 cores on one chip and not one die, just a typo sorry for everyone that I confused with that. But all your doing is trying to complicate things by saying it differently, yes MS considers it one physical processor but that does not mean that it is one core running six threads. So you dont think I'm trying to repeat what you said the difference unless you worded it wrong or I didn't interepret what you said in the way you meant is that it is essentially 3 processors running 2 threads each not one running 6 threads by virtually splitting and processing threads in pairs. But you are right when you say its physically one all the cores "are" on the same chip but to anyone that doesn't speak techo babble they should no it is 3 "pairs" of threads being calculated at 3.2 ghz each
Fine, you made your point, but the debate is not over. There is no "right" or "wrong" interpretation to a multi-core single-packaged processing device being one processor or multiple processors. Whether you encase single-core processing devices into individual, separate packages that communicate over an external bus or if you integrate those single-core processing devices into a single multi-core processing device (where the cores communicate through an internal bus as with the Athlon64 X2 or with an external bus as with the first crop of Pentium Ds) you still have multiple "cores". The distinction of single-processor versus multiple-processor is, at this time, primarily a legal one. And laws change.Telomar said:
still so wrong, so wrong
Please, then explain the practical difference between:
I believe, from the programmer's and operatings system's viewpoint, that the answer is "none".
From the view of the software, a "core" is fully functional independent processor. Both logically and behaviourally, a "core"is a processor.
A "thread" (as in "SMT" or "HyperThreading"), however, appears logically as a unique processor - but behaviourally it is clear that it isn't a fully independent processor but is sharing execution units with other logical processors.
Your link to Microsoft talks about licensing issues. Microsoft has decided to license dual-core system as single processors. This has absolutely no relevance to a technical debate - it's purely a business model issue!
Should I send you pointers to Oracle's pages about how they will be licensing dual core as dual processor?
Telomar said:
Please, then explain the practical difference between:
- Three cores on one chip
- Three chips with one core each
I believe, from the programmer's and operatings system's viewpoint, that the answer is "none".
From the view of the software, a "core" is fully functional independent processor. Both logically and behaviourally, a "core"is a processor.
A "thread" (as in "SMT" or "HyperThreading"), however, appears logically as a unique processor - but behaviourally it is clear that it isn't a fully independent processor but is sharing execution units with other logical processors.
Your link to Microsoft talks about licensing issues. Microsoft has decided to license dual-core system as single processors. This has absolutely no relevance to a technical debate - it's purely a business model issue!
Should I send you pointers to Oracle's pages about how they will be licensing dual core as dual processor?
ksz said:
True.
It only becomes "right" or "wrong" when you try to include behaviour in the equation.
Behaviourally, the system will have a very hard time distinguishing between a dual core chip and a system with dual single core chips.
Behaviourally, it is pretty easy to distinguish a dual-thread SMT (hyperthread) chip from a dual-core system or a dual single-core system.
On the non-technical front, things get very complicated.
Windows XP Home doesn't support dual processors, but it runs in SMP mode to support a hyperthreaded CPU.
XP Pro supports dual processors (not more) - but if you have dual HT CPUs it says that you have a quad processor system.
For most people, it's pretty clear that
- a "core" as in "dual-core" is a full processor
- a thread (as in SMT or HyperThreading) is logically a unique processor, but behaviourally it shares resources and execution units with other logical processors and is less than a full CPU
Whether or not a particular application or OS decides to license a dual-core CPU as "one processor" or "two processors" is a business call, not a technical issue. The OS will report the actual topology to the application -- the folks who implement the licensing policy for the app or OS decide whether or not the configuration is supported.
bushgreen said:
Your forgetting that the Xbox does not have G5's!
Did you read the first few pages of this thread?
iGary said:
what's to say that Apple doesn't have the latest procs?
whether Apple chooses to say anything about them or chooses to use them or not is another matter entirely...