Xp vs Vista, and processor questions...
Discussion
UKbob said:
It is, however, THE most appallingly pathetic OS in terms of stability, ever. Not that anyone here didnt know that, but it really is terrible. Vista itself has never once crashed, in fact it seems rather good at not falling over itself. But the applications all do, and frequently. Outlook up to 5 times a day. So far more frequently than ME as I recall, its one hell of a bastard of an operating system in that respect, which makes it a bit strange. All the more reason to abandon ship.
Odd - I've been running it every day on my main machines for over a year now, and haven't had the problems you're experiencing. Pretty sure there are others on here in the same boat, so I'd say there's something wrong with yours - not that there's something it.Saying "blummin' Vista, pile of rubbish" is easy, but in your position I might be more inclined to work out what was actually wrong.
jamieboy said:
Odd - I've been running it every day on my main machines for over a year now, and haven't had the problems you're experiencing. Pretty sure there are others on here in the same boat, so I'd say there's something wrong with yours - not that there's something it.
Saying "blummin' Vista, pile of rubbish" is easy, but in your position I might be more inclined to work out what was actually wrong.
I agree. I have no problems with crashing apps under Vista. It's all rather pleasant in fact and just works. And has been for the past year or so.Saying "blummin' Vista, pile of rubbish" is easy, but in your position I might be more inclined to work out what was actually wrong.
Zod said:
I own, inter alia, a 3.4GHz P4 PC and a Q6600 PC and I can tell the Learner that the Q6600 is significantly faster for every task, whether gaming, video or office.
Inter alia? Feh - my Mac Pro (with some Intel stuff in it, apparently called Xeon E5462) is proper fast, and after only a week it's 11th in the world at that Seti thing, beating all Q type CPUs. The only thing above it that isn't another Mac Pro or a Windows Server running the same Xeon chip is some overclocker with a quad-core Opteron 8347 on Linux.I've always (been an Apple fan) loved the 'difference' of the PowerPC architecture and the 'better' efficiency of the architecture, yadda yadda. But Intel have either just been incredible engineers and called their chips 'x86 compatible', or they've triumphed with development over design. Their CPUs are a LOT faster than Apples with previous CPUs. Simple as that, I can't deny it.
That said, OS X does use multiple CPUs pretty well (Apple sold dual-CPU workstations as a matter of course well before they were standard-issue for PC users) so perhaps that'd explain the reason why Apple do well with the new multi-core chips.
I certainly wouldn't be that impressed with an 8-core system running a single-thread app on a 32-bit operating system with poor thread support
For multitasking with corporate workflows, multiple cores and lots of RAM win big style - run everything at the same time. For the gamer overclocker community, they're less relevant unless your game engine is multithreaded (and the graphics libraries are similarly multithreaded and thread-safe). That was always the argument - a fast single CPU was always best for games, due to OpenGL / Direct3D libraries being singlethreaded and the game engines being singlethreaded. Now that's changing, multiple cores can only be a good thing.
As long as the northbridge is up to the job of dealing with 8+ cores asking for memory access!!!!! That's the real bottom line unless your algorithms fit into CPU cache, and with Intel's quadcore CPUs, they don't seem to have a dedicated cache per core, unless someone can correct me...
AMD's integrated memory controllers could give them an advantage here - the Opteron was faster than the G5 (only usable 64-bit CPUs at the time in 2003) because of the Opteron's integrated RAM controller - the G5 had some crazy northbridge that was massively overworked (I was seeing temps of 110+ deg C under load on my early example...)
But I know nothing about fast PC stuff when it comes to Windows... a decent OS though will use the multiple cores (assuming you have extra memory to suit) and make everything move faster. A pure gamer will disagree. For general use though, the multiple cores will win. Just make sure that you increase RAM in line with the amount of apps you use. Having a quad-core CPU that happily runs 3 of your main apps concurrently means you'll need to spec the box to have the amount of RAM you'd use to run all three concurrently. I/O is always the bottleneck, and super-duper CPUs push you into using swap earlier and harder. Buy more RAM!!!!!
cyberface said:
Inter alia? Feh - my Mac Pro (with some Intel stuff in it, apparently called Xeon E5462) is proper fast, and after only a week it's 11th in the world at that Seti thing, beating all Q type CPUs. The only thing above it that isn't another Mac Pro or a Windows Server running the same Xeon chip is some overclocker with a quad-core Opteron 8347 on Linux.
AMD's integrated memory controllers could give them an advantage here - the Opteron was faster than the G5 (only usable 64-bit CPUs at the time in 2003) because of the Opteron's integrated RAM controller - the G5 had some crazy northbridge that was massively overworked (I was seeing temps of 110+ deg C under load on my early example...)
I'd never heard of this Seti thing until you mentioned it, be interesting to see how my work PC compared to it (QX6850 quad-core @ 3.0GHz) there is a recent revision to it, but that is the fastest desktop CPU intel make. It's a bit strapped on RAM at the moment, so I think i'll suggest upgrading that AMD's integrated memory controllers could give them an advantage here - the Opteron was faster than the G5 (only usable 64-bit CPUs at the time in 2003) because of the Opteron's integrated RAM controller - the G5 had some crazy northbridge that was massively overworked (I was seeing temps of 110+ deg C under load on my early example...)
The AMD IMC has always been a benefit in virtualisation and synthetic memory tests & benchmarks, until Intel have it on their CPU's, which is planned for their "Nehalem" architecture, they will always lag behind AMD in memory throughput.
Interestingly the aforementioned Nehalem will pretty much integrate all of AMD's developments (HT not FSB, IMC and native quad-core on a 45nm die) onto an Intel chip.
LukeBird said:
cyberface said:
Inter alia? Feh - my Mac Pro (with some Intel stuff in it, apparently called Xeon E5462) is proper fast, and after only a week it's 11th in the world at that Seti thing, beating all Q type CPUs. The only thing above it that isn't another Mac Pro or a Windows Server running the same Xeon chip is some overclocker with a quad-core Opteron 8347 on Linux.
AMD's integrated memory controllers could give them an advantage here - the Opteron was faster than the G5 (only usable 64-bit CPUs at the time in 2003) because of the Opteron's integrated RAM controller - the G5 had some crazy northbridge that was massively overworked (I was seeing temps of 110+ deg C under load on my early example...)
I'd never heard of this Seti thing until you mentioned it, be interesting to see how my work PC compared to it (QX6850 quad-core @ 3.0GHz) there is a recent revision to it, but that is the fastest desktop CPU intel make. It's a bit strapped on RAM at the moment, so I think i'll suggest upgrading that AMD's integrated memory controllers could give them an advantage here - the Opteron was faster than the G5 (only usable 64-bit CPUs at the time in 2003) because of the Opteron's integrated RAM controller - the G5 had some crazy northbridge that was massively overworked (I was seeing temps of 110+ deg C under load on my early example...)
The AMD IMC has always been a benefit in virtualisation and synthetic memory tests & benchmarks, until Intel have it on their CPU's, which is planned for their "Nehalem" architecture, they will always lag behind AMD in memory throughput.
Interestingly the aforementioned Nehalem will pretty much integrate all of AMD's developments (HT not FSB, IMC and native quad-core on a 45nm die) onto an Intel chip.
The reason why I like the Seti thing is that it's actually a *real* scientific project - the actual algorithms are optimisable for each platform's strengths (the reason why my Quad G5 was #1 on the planet a few years ago) but it differs from synthetic benchmarks in one key fashion - it's doing *real scientific computing* and the 'geek dick-swinging contest' is a secondary factor.
If anything, the project organisers have been very clever putting leaderboards and teams and 'who's got the fastest computer' rubbish up, because it has tapped into the competitive spirit of geeks everywhere. It's a good way to see whether your water-cooled, massively-overclocked, custom rig holds together under real-world heavy scientific processing. Or whether my (relatively) cheap off-the-shelf 'ooh Apple are such ripoff merchants' box actually runs fast.
On top of that, it's doing something of value to mankind. They were the first - the same concept (distributed processing, broken into teams and individuals competing to process the most work units) has been spread out onto other scientific endeavours, such as climate prediction, malaria control, and a host of mathematical / astronomical calculations that rely on vast searches (prime number stuff etc.). Each project will have its own algorithm - for example a prime number searcher will most likely be an integer maths job and Intel chips have always been fastest at that. The Seti project needed loads of fast floating point maths (fourier transforms, mostly) and that's where the Altivec'd G5s shat over everyone.
Moral of this story? Apple Mac Pros are *astonishingly* impressive value cost-per-performance at the moment. Since the boxes are so nicely engineered and made of proper metal... and there appears to be no cost-cutting inside (pluggable drives, proper fans, nothing cheap-looking inside) - Apple must be getting some crazy sweetheart deal from Intel on their top-end chips, otherwise Dell should be able to sell an equivalent machine for much less... but they can't.
Cozying up with Otellini might have been one of the wisest moves Jobs did... if anyone remembers the keynote when Apple announced they were switching to Intel CPUs, Otellini ran onto the stage and
ing HUGGED teh Steve... he's either one hell of an actor or he was well chuffed to be able to sell chips into a new market unafflicted by the stagnation and rush-to-the-bottom of the so-called 'Wintel' market. Yet either he's selling his best chips at a fat discount to Apple, or there's some other chicanery going on. Anyway this is well off topic. (PS. I'll be 9th in the world on next refresh - takes a while for the central computers to ensure you're not cheating)
cyberface said:
Zod said:
I own, inter alia, a 3.4GHz P4 PC and a Q6600 PC and I can tell the Learner that the Q6600 is significantly faster for every task, whether gaming, video or office.
Inter alia? Feh - my Mac Pro (with some Intel stuff in it, apparently called Xeon E5462) is proper fast, and after only a week it's 11th in the world at that Seti thing, beating all Q type CPUs. The only thing above it that isn't another Mac Pro or a Windows Server running the same Xeon chip is some overclocker with a quad-core Opteron 8347 on Linux.I've always (been an Apple fan) loved the 'difference' of the PowerPC architecture and the 'better' efficiency of the architecture, yadda yadda. But Intel have either just been incredible engineers and called their chips 'x86 compatible', or they've triumphed with development over design. Their CPUs are a LOT faster than Apples with previous CPUs. Simple as that, I can't deny it.
That said, OS X does use multiple CPUs pretty well (Apple sold dual-CPU workstations as a matter of course well before they were standard-issue for PC users) so perhaps that'd explain the reason why Apple do well with the new multi-core chips.
I certainly wouldn't be that impressed with an 8-core system running a single-thread app on a 32-bit operating system with poor thread support
For multitasking with corporate workflows, multiple cores and lots of RAM win big style - run everything at the same time. For the gamer overclocker community, they're less relevant unless your game engine is multithreaded (and the graphics libraries are similarly multithreaded and thread-safe). That was always the argument - a fast single CPU was always best for games, due to OpenGL / Direct3D libraries being singlethreaded and the game engines being singlethreaded. Now that's changing, multiple cores can only be a good thing.
As long as the northbridge is up to the job of dealing with 8+ cores asking for memory access!!!!! That's the real bottom line unless your algorithms fit into CPU cache, and with Intel's quadcore CPUs, they don't seem to have a dedicated cache per core, unless someone can correct me...
AMD's integrated memory controllers could give them an advantage here - the Opteron was faster than the G5 (only usable 64-bit CPUs at the time in 2003) because of the Opteron's integrated RAM controller - the G5 had some crazy northbridge that was massively overworked (I was seeing temps of 110+ deg C under load on my early example...)
But I know nothing about fast PC stuff when it comes to Windows... a decent OS though will use the multiple cores (assuming you have extra memory to suit) and make everything move faster. A pure gamer will disagree. For general use though, the multiple cores will win. Just make sure that you increase RAM in line with the amount of apps you use. Having a quad-core CPU that happily runs 3 of your main apps concurrently means you'll need to spec the box to have the amount of RAM you'd use to run all three concurrently. I/O is always the bottleneck, and super-duper CPUs push you into using swap earlier and harder. Buy more RAM!!!!!
2. I also have two PCs with AMD processors, a 4400 X2 and a 4600 X2. They are much slower than the Q6600. AMD should be able to be quicker with their integrated memory controller (as they were two years ago), but even their new Phenom range is way, way behind intel's products.
3. The latest games use multiple cores.
Morningside said:
Someone told me that Vista made better use of the Dualcore/processor than what XP ever did and even running a single processor program it will try to multi-tread it and make it work better - I cannot believe this to be true?
What they probably mean is that they have made more of the Windows APIs thread-safe, so any ostentisbly 'single-threaded' app that calls Windows routines can have the Windows routines be scheduled onto another core by the kernel.If you've written an algorithm that is single-threaded by nature and serialised, and doesn't call anything outside your own code, then you're right - no OS is going to be able to magically multi-thread it for you...
cyberface said:
Morningside said:
Someone told me that Vista made better use of the Dualcore/processor than what XP ever did and even running a single processor program it will try to multi-tread it and make it work better - I cannot believe this to be true?
What they probably mean is that they have made more of the Windows APIs thread-safe, so any ostentisbly 'single-threaded' app that calls Windows routines can have the Windows routines be scheduled onto another core by the kernel.If you've written an algorithm that is single-threaded by nature and serialised, and doesn't call anything outside your own code, then you're right - no OS is going to be able to magically multi-thread it for you...
cyberface said:
The reason why I like the Seti thing is that it's actually a *real* scientific project - the actual algorithms are optimisable for each platform's strengths (the reason why my Quad G5 was #1 on the planet a few years ago) but it differs from synthetic benchmarks in one key fashion - it's doing *real scientific computing* and the 'geek dick-swinging contest' is a secondary factor.
If anything, the project organisers have been very clever putting leaderboards and teams and 'who's got the fastest computer' rubbish up, because it has tapped into the competitive spirit of geeks everywhere. It's a good way to see whether your water-cooled, massively-overclocked, custom rig holds together under real-world heavy scientific processing. Or whether my (relatively) cheap off-the-shelf 'ooh Apple are such ripoff merchants' box actually runs fast.
I did have a look at the website, I see what you mean, Xeon's rule supreme! If anything, the project organisers have been very clever putting leaderboards and teams and 'who's got the fastest computer' rubbish up, because it has tapped into the competitive spirit of geeks everywhere. It's a good way to see whether your water-cooled, massively-overclocked, custom rig holds together under real-world heavy scientific processing. Or whether my (relatively) cheap off-the-shelf 'ooh Apple are such ripoff merchants' box actually runs fast.
When I get my PC back together I'll have to try it on there to see how it comes out. Very, very low, but as you said, it's cock-waving & helping!
Be interesting to see if anyone is running a system like mine on there!
I assume it only really stresses CPU & RAM?
I believe Intel's next generation processors will do away with the northbridge as we know it.
Currently the 45nm processors aren't much faster than the old ones so I'm expecting a big hike in performance in the not too distant future with Nehalem which is an entirely new architecture. Is anyone else close to having a workable 45nm process?
http://en.wikipedia.org/wiki/Nehalem_(CPU_architec...
Currently the 45nm processors aren't much faster than the old ones so I'm expecting a big hike in performance in the not too distant future with Nehalem which is an entirely new architecture. Is anyone else close to having a workable 45nm process?
http://en.wikipedia.org/wiki/Nehalem_(CPU_architec...
Edited by mackie1 on Monday 4th February 11:16
45nm Yorkfield quads have been delayed because of CPU errata, so perhaps Intel's 45nm process isn't quite as spanktastic as first envisaged. Personally, I don't think it offers enough over their (very good) 65nm to warrant a change. I do like my AMD stuff, but even I can admit that! 
AMD are apparently on target with 45nm stuff, and there are rumours that they'll push dev on it to get past their disappointing 65nm Phenom (once they get a decent one over 2.6Ghz I'll swap one for my 6000+)unless, the B3 stepping that'll cure the current errata that is restricting clock speed is, exceptionally good and makes up the some of the difference that Intel have pulled ahead.
Sorry if thats incoherent, I'm in a rush!
AMD are apparently on target with 45nm stuff, and there are rumours that they'll push dev on it to get past their disappointing 65nm Phenom (once they get a decent one over 2.6Ghz I'll swap one for my 6000+)unless, the B3 stepping that'll cure the current errata that is restricting clock speed is, exceptionally good and makes up the some of the difference that Intel have pulled ahead.
Sorry if thats incoherent, I'm in a rush!
anonymous said:
[redacted]
As the first para says, all internet bandying!Be interesting to see if Intel do announce anything.
As I said, I don't think 45nm cores really brought all the benefits people were hoping for.
To be honest as a gamer, the difference between my 6000+ and any new CPU would be negligible at best.
Photoshop & Premiere Pro are more than quick enough for rendering & filtering for me as well
Interesting article though, whoever wrote it
anonymous said:
[redacted]
I'm interested in your '1 GHz of overclocking headroom' comment.I'm maxing my 8 cores on my Mac Pro (base model, 2.8 GHz clock) on that Seti thing 24 hours a day. The fans are resolutely stuck to their minimum rotational speed... this says that the entire package is either monstrously overcooled, or I can rack the chips up a hell of a lot.
Now the question is - how do you overclock an EFI-based Intel Xeon system running OS X???
bo
ocks. I gave all this PC overclock shit up years ago when I moved over to running OS X. Damn Intel's got me back again. I mean - 2.8 GHz at full pelt, all 8 cores slamming away, and it's not even getting warm??? Got to be able to rack them up to 3.4 surely???? anonymous said:
[redacted]
Surely that can't be the case, any current game is going to be GPU, not CPU bound.Widely regard that a 2.4Ghz or so Core 2 or equivalent is going to be more than enough for any current game.
Upgrading from an old single-core 3700+ on Skt754 to my 6000+ yielded little benefit in any games (I kept the same video card for a month or so). FPS was upped in several games (mostly source based) which i guess is the CPU handling some of the in game physics and the like.
cyberface said:
anonymous said:
[redacted]
I'm interested in your '1 GHz of overclocking headroom' comment.I'm maxing my 8 cores on my Mac Pro (base model, 2.8 GHz clock) on that Seti thing 24 hours a day. The fans are resolutely stuck to their minimum rotational speed... this says that the entire package is either monstrously overcooled, or I can rack the chips up a hell of a lot.
Now the question is - how do you overclock an EFI-based Intel Xeon system running OS X???
bo
ocks. I gave all this PC overclock shit up years ago when I moved over to running OS X. Damn Intel's got me back again. I mean - 2.8 GHz at full pelt, all 8 cores slamming away, and it's not even getting warm??? Got to be able to rack them up to 3.4 surely???? http://en.wikipedia.org/wiki/Intel_SkullTrail
Article on crazy Intel Skulltrail system said:
Skulltrail has a front side bus rate of 1600 MHz, and it was demonstrated with two 45 nanometer High-K processors running at 3.2 GHz. During IDF, a 4.0 GHz phase cooled Skulltrail system was demonstrated. Then on 22 October 2007, the two processors were demonstrated running at 4.4 GHz, water cooled. They were demonstrated again on October 31 2007, this time running at 5.0 GHz, phase cooled. The 1st benchmark of Skulltrail overclocked showed up 13th of December 2007, and broke the record for Cinebench 10 rendering speed
In essence, I'd guess the CPU's are fairly similar to the Xeon's in your Mac Pro. Quite how you go about OCing them in OS X is a bit of a mystery though!
anonymous said:
[redacted]
Cyberface has one of the new Mac Pro's so they are the new 45nm Xeon's, whatever the core is called... anonymous said:
[redacted]
I reckon I'll go quad when there is a Phenom B3 at 2.6Ghz+ or so. Read the Tom's hardware review on Skulltrail today, personally I think it'll be as successful as AMD's 4X4, i.e. too power hungry, noisy and too server orientated to take advantage of all the power, FB-DIMMs and the like. Impressive none the less!
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff