Hi
I intend to buy a new PC, Are these hardware compatible with Linux(Debian) ?

CPU : INTEL I5 750
M.B : Gigabyte P55-UD3P
RAM : 2X2GIG CORSAIR DDR3 1333
HDD : SEAGATE 500
VGA : GEFORCE 8400 256MB DDR2

Thanks in advance

Should be OK
However, most problems are with wireless cards, webcams, networkcards.....

More info
http://tldp.org/HOWTO/Hardware-HOWTO/
http://www.linux-drivers.org/
http://www.linux.org/hardware/

Thanks for your attention
Generally the kernel supports most of network cards.nic of this mainboard is on board with Realtalk chip,Do you think kernel does not support this chip??
Thanks in advance

The Gigabyte and JMicron SATA controllers will be a problem, but the NIC should work.

To correct what repo is trying to say. Linux has trouble with the extras that are on the motherboard such as NIC, sound, storage controllers, sound, USB, IEEE-1394 (aka Firewire or i.Link), WiFi, audio amplifier, and graphics.

Thanks for your attention
In wikipeida is mentioned that "The most recent Linux kernel and the most recent JMicron controller BIOS[13] solved these incompatibilities, but may not be present in existing products, or may require re-flashing the motherboard BIOS", Is that true? That means there is way to solve this problem in kernel ?
In specification of this motherboard, I have read 3 chip for SATA controller,"Intel® P55 Express","JMB363" and "GigaByte", Is there any problem with Intel® P55 Express cip?
May you suggest some motherboards with full Linux compatibility ?in addition to I intend buy Core I5 750, So my selections are limited, all right ?
About VGA card, Is there any problem with Linux ??

Thanks in advance

If you haven't purchased the mobo yet ... I'd look at a different mobo, the ASUS & EVGA boards look nice. I have had lackluster performance and compatibility issue's with JMicron.

This is the board I would purchase myself. BTW your processor shouldn't be an issue, Nor should your Nvidia card, personally I would get a newer card, like the Nvidia GT 220.

http://www.newegg.com/Product/Produc...82E16813131604

Thanks for your reply
About VGA, My budget does not allow to buy higher VGA card
Regarding to motherboard, "Asus P7P55-LX" has SATA chip Intel® P55 Express Chipset as the same as "Gigabyte P55-UD3P", then should not any problem with GigaByte!!that's right??
Thanks for more suggestions

The reason why I did not say anything about the video card is that it is easy to look that up by visiting nVidia's web site and check the support list. It will take you about 5 minutes. Since you are a Debian user, I assume you have the knowledge to look up something your self. Also if you are picking a nVidia graphics card, it is a given that is already supported in Linux.

The primary SATA controller should work, so do not make a big deal about it. If you are making a big deal go with AMD systems because they are already known to work out of the box in Linux.

If you are on a tight budget an AMD will be a better buy. An AMD Phenom II X4 945 has the same performance as the Intel i5 750 and Intel i7 920 at a lower price. You will save over 30 US dollars by going with AMD compared to Intel for the processors. For the motherboard you will save over 50 US dollars. Sure you can go that pristine computer setup or you can go with a equal setup and save money.


AMD Phenom II X4 945 $165.99
BIOSTAR TForce TA790GX A3+ $109.99
2 X Crucial Ballistix 2GB (2 x 1GB) DDR3 1333 BL2KIT12864BA1336 $93.98
Total: $369.96

or

AMD Phenom II X4 945 $165.99
MSI 790FX-GD70 $163.99
2 X Crucial Ballistix 2GB (2 x 1GB) DDR3 1333 BL2KIT12864BA1336 $93.98
Total: $423.96

or

Intel i5 750 $199.99
GIGABYTE GA-P55-UD3P ~$185
2 X Crucial Ballistix 2GB (2 x 1GB) DDR3 1333 BL2KIT12864BA1336 $93.98
Total: $478.97

I am comparing the prices of AMD and Intel systems. AMD gives you more features for less money. If you are comparing power consumption between two systems, they are about equal on a total power consumption. The only difference here is my wallet and your wallet. Your not paying the money with my wallet because you are paying the system with your wallet. I prefer AMD because I know they are better in real world applications compared to what the benchmarks are showing. If you do not have a budget, go with Intel, but it is nice to buy something cheaper when something similar are equal.

thanks for your attention
About comparison of "Core I5 750" and "Phenom II 965" I found this comparison, Do you have any suggestion about this comparison ??

http://www.tomshardware.com/reviews/...65,2389-2.html

Thanks in advance

Be careful comparing an AMD Phenom II to Intel i5 or i7 because not all benchmark sites set controls. The Intel i5 and i7 includes the gimmick Hyperthreading and Turbo. If you understand these features, these are not fair to use during benchmarking or comparing. The feature Turbo is just over clocking the processor to provide a feeling that the processor is gaining performance at a cost of damaging the processor. Also Hyperthreading doubles the amount of processors which helps a lot when multi-threading programs are used. I doubt tomshardware set controls, so Turbo and Hyperthreading are set to enabled for their benchmarks. Also those programs that tomshardware are showing are not native to Linux, so the benchmark is useless. One thing comes to mind it is stating simulated this means it is what the program thinks how the processor will perform. Thinking and realistically are two different things. The following is another reviewer and it is probably more accurate.

http://www.hardwaresecrets.com/article/819/

I do not know how they calculate performance difference or how they did the benchmark, so results are still irrelevant. The performance difference formula that I use is ((X - Y)/(X + Y)) * 100. I would only say one processor is faster than the other is when it is at least 10% or faster. Of course it is best to find something that is 200% faster.

With out knowing what you are going to do with your setup. At this time, I still suggest an AMD Phenom II X4 945 since most of your parts are not high performance and just picking the fastest processor does not help at all to makes things faster. You could go an even cheaper processor like a Athlon II X4 620 or Athlon II X2 250.

Thanks for your attention
I will use this system for virtualization based on Xen technology, Is AMD compatible with Xen technology ??

Thanks in advance

And if these features speed up performance, what is wrong with including them in testing?

There hasn't been a good Intel scandal for years and years, so I would be interested in any evidence that anyone who has stuck to the design guides and hasn't separately overclocked their processor has suffered damage.

Yes, that can be true, but if it does, what is wrong with including that in the benchmarking? Would you include it if it slowed down the processor?

Not useless, but of limited usefulness, certainly. But you need to use some intelligence to compare any benchmarks that aren't benchmarks of exactly the program or program combination that you intend to use under exactly the conditions that you intend to use it/them, so that is only a matter of degree.

Usually when tomshardware does 'simulated' they mean that they didn't have the exact processor to be tested, but they have adjusted the processor clock and the memory clock in order to give it the same performance as the processor that they don't have available. This process is usually quite accurate, within a few percent, so

seems excessive. No one has just 'thought' what the new processor will do, they have tested the closest analogue they can get to that processor. Particularly when you can't expect general benchmarking to tell you anything within a percent or two.

Now that is true. You don't have to go all that far up the the price range for Intel to be a clear winner, but down at the low/mid end AMD does win, even if only on motherboard costs. Whether four cores helps, or not, is an open question. Getting two higher clocked cores for the same or a lower price is a better deal if you can't make use of four cores and if you can't make use of four cores, eight virtual cores isn't going to be of much help either.

Yes, but you really need to do research because this information is everywhere.

I prefer VMware because it is more reliable and stable. Also VMware is easier to setup.



Using Hyperthreading does change the tables how people read the results. There is one processor that has eight processors and other processor has four. If the program being used is multi-threaded or if the OS is able to balance the the processors equally, the setup with the most processors will win. Over clocking also changes the tables. If processor A is running at 3 GHz and processor B is running at 4 GHz which is an over clocked 2.6 GHz where do you think will win the tests. Processor B will win the tests.

A true benchmark is knowing what changes the levels of the field, so they are not included.



There have been many scandals from Intel. Also the EU is not giving up with the lawsuit with Intel about Intel black mailing or bribing companies to only sell Intel products.


I am not the one that should be told to have intelligence. It seems you should rephrase to you that get some intelligence.

Including any simulation results into benchmarks is just stupid. Simulations are what they are simulations, so they should never be looked at when relating to real world tests. The only way simulations can have big impact is if the simulations has enough information to use during simulations. Since tomshardware.com has changed their site, their benchmarks have changed to be questionable. Their benchmarks are not correct and misleading. At least anandtech.com takes the time to do the benchmarks correctly by keeping the benchmarks on a level plain. They state if they are using Hyperthreading and are using Turbo since they that AMD does not have these features and it creates an unfair advantage that can hurt anandtech's reputation if they do not state it.


Again you are the only one that comes here saying that four processors can not be loaded with out multi-threading programs. Yes, this is true in the Windows world. Linux has a different story. The Linux kernel tries to load each processor in the system evenly. It tries so well that it is almost perfect. I recommend go with computers with the most amount of processors. The speed of the processor in Linux is irrelevant since Linux is a multi-tasking operating system. A multi-tasking operating system sets it self up by creating virtual environments for each program, so the kernel can organize what program or environment can load on what processor. Windows is a multi-processing operating system, so it is limited to using one processor until a multi-threaded program is run. A multi-processing operating system loads one environment that gets shared.

Are you complaining that someone might read a set of benchmark results and unfairly see which one comes out as the fastest?

That is far from necessarily true. There is a memory interface and you can't get more information through the memory than will go through the memory interface. As you point out in your next point, clock rates are important, cache, instructions per clock. So, just simplifying things to 'its only the number of cores' is not all that sensible.

Well, at least you have admitted that clock rate has something to do with it, even if you have managed to contradict your previous point. Of course, you are also being quite ridiculous, because no one is overclocking a a 2.6 GHz processor to 4 GHz, or anything like.

An interesting definition of a benchmark. Most people would consider that you have to be clear and upfront about what you are testing, and you haven't made any criticism on that front, but clearly you don't like the result.


...good scandals, like the pentium floating point bug... the issues that you mention are commercial issues, and while you may or may not think having these is a bad thing, it doesn't have that much to do with the the processors' performance in benchmarks.

If you want to make the case that Intel indulges in commercially underhand business practices, say that and make that case so we can all discuss that rationally rather than disguise it as an argument about benchmarking. And the OP has the chance to decide how much weight it should be given, bearing in mind their own circumstances.

I notice that you haven't come up with the evidence of processor damage, so should we assume that so I would be interested in any evidence that you haven't found a single case of someone who has stuck to the design guides and hasn't separately overclocked their processor and damaged the processor?


You seem touchy about that. Given that there was a general comment about the nature of benchmarking and what was needed to interpret them, choosing to see it as a personal insult then and choosing to resort to personal insults in reply is gong some (and against the rules of this forum, btw).

Or, have you proved so poor at expressing your position that I have made an unfortunate mis-interpretation of your position.

Simulations can be accurate or inaccurate. You don't make the case that they are inaccurate in any way that the rest of us can test (unlike tomshardware, who, in all the benchmarks that I have seen, publish exactly the conditions in which the benchmarks are performed).

Literally, all benchmarks are questionable. That's why you have to use intelligence to understand and apply them. Please re-read the comment on intelligence and try to understand it this time.

This is like saying 'these AMD processors don't have level 3 cache, so we'll have to turn off the level 3 cache on the Intel ones to make it a fair test'. The processors have the features that they have, and the features that speed up the processor, speed up the processor, whether that's cache clock speed or whatever. That's what was wrong with your '2.6GHz processor overclocked to 4 GHz' analogy; that isn't what happening, the processor isn't overclocked to 4 GHz.

It would be different if you had been able to make the case that loads of people were having their processor fail because of the 'slight 'auto-overclocking' feature; then you could have argued that it was illegitimate because you dare not really turn the feature on in practice, so you had to keep it off in benchmarking.

Please point out to me where I say, in this context, "that four processors can not be loaded with out multi-threading programs" because I seem to have missed it. It is certainly possible in a server context to have enough programs running to load multiple cpus, real or virtual, but it is unclear how that relates to the situation of the OP. Perhaps I am wrong in believing the OP is a desktop user, but that was the interpretation that I made from the little evidence available.

Unfortunately, this is far from true on the desktop, in the most usual case. The kernel may do as good a job as it can, but without many independent programs (thus avoiding the data dependency problem) or tasks written with parallel execution in mind specifically, the kernel itself can't do much about it. The fact is, for these load cases, there is often no mathematical combination of the instantaneous loads to result in even loading, and a combination of the average loads for even loading does not necessarily produce even loading at any particular time.

(Oh, and by the way, this 'even loading' is irrelevant, except in a job-queued environment, which may be true of IO, depending on which scheduler you use (see, for example, BFS!) and may be true for certain types of server, but is only true to a very limited extent for a desktop user with an interactive task load. And it may even result in increased power dissipation, which is not an advantage.)

It is difficult to see a consistent pattern of advice emerging here; it would be bad to use an Intel CPU, because they use hyperthreading, which doubles the number of processors, and doubling the number of processors is somehow cheating, but you should choose the maximum number of processors.

And, obviously, now that you have told me that the speed of the processor is irrelevant, I'll get out my old K-6 400, as you claim that must be as fast as a more modern processor on Linux (clue: it isn't).

BTW, I notice that you didn't react to "Would you include it if it slowed down the processor?" Given that there are known to be use cases in which hyperthreading does slow down the processor, I don't know why you failed to take that opportunity.

Bury the axe of war, dudes. =) One will buy Pentium, one AMD, it was always like that, and i wish it will remain so.