CPU comparisons (I need your help)

  • Thread starter The_Absolute
  • Start date
  • Tags
    cpu
In summary, the conversation discussed the speaker's old desktop PC that had stopped working, and their plans to get a new one as a Christmas gift. They were seeking a professional opinion on which processor would be better for the new desktop, comparing an AMD Phenom X4 with 8 GBs of RAM to an Intel Core 2 Quad with 8 GBs of RAM. The conversation also touched on the difference in computing power between single core processors and multi-core processors, as well as the impact of clock speeds and cores on performance. The speaker also asked about the possibility of overclocking their new CPU and the potential benefits of upgrading to a better graphics card. It was noted that Windows Vista can use a significant amount of RAM, and there
  • #1
The_Absolute
174
0
My old desktop PC which I built circa 2003, has officially stopped working. The parts in it are discontinued and I can't find them to buy anywhere. I am getting a new desktop as a christmas gift. I wanted you to tell me your professional opinion on which processor is better.

I'm buying a desktop with an AMD Phenom X4 2.2 GHz with 8 GBs of RAM. I looked at another computer which had an Intel Core 2 Quad 2.3 GHz with 8 GBs RAM. Both had windows vista 64-bit OS installed on them. My old PC that I'm practically forced to throw away, besides a few parts on it like the graphics card, memory, drives, sound card, and PSU that I may be able to get about $150-200 for if I sell them all, was running a Pentium 4 HT 3.0 GHz, 1 GB of RAM, and an ATI Radeon X800 Pro 256 MB. If you compare a single core of the 2.2 GHz Phenom, and the 2.3 GHz Core 2, do their slightly lower clock speeds really give it less computing power than opposed to the Pentium 4 3.0 GHz? Or the Core 2's and Phenoms that par, or suprass 3.0 GHz?

How much more computing power does having a quad-core CPU with higher clock speeds, such as 2.2 GHz as opposed to 3.3++ give you? Is a single core of the Phenom or Core 2 more powerful than a Pentium 4 with a 3.0 GHz clock? I've been told that clock speeds aren't everything when it comes to processors, especially multi-core designs. I've heard of today's latest quad-core processors overclocked to 4.0 GHz. And obviously it's going to be a lot faster with 8 gigabytes of ram as opposed to only one.

A little bit off-topic, but the computer I think I'm getting has a Radeon HD 3100 in it, which is underpowered from what I've been told. At least when it comes to running graphics intensive programs such as Crysis and any other DX10 game. Is the Radeon HD 3100 a lot less powerful than the X800 Pro? Will I still be able to run DX9 games such as Battlefield 2 on it? If I were to upgrade to a better GPU, such as an Nvidia 8800-9800 GTS or GTX, would I have to buy a new power supply? I think the computer I'm getting only has a 400 watt in it. At full load, the high-end cards can consume 400++ watts.
 
Technology news on Phys.org
  • #2
The_Absolute said:
How much more computing power does having a quad-core CPU with higher clock speeds, such as 2.2 GHz as opposed to 3.3++ give you?

Dude, this doesn't make any sense. Computing power difference is easy to find, 3.3 GHz - 2.2 GHz = 1.1 GHz, so the later has that much more computing power.

Issues with cores and symmetric multiprocessing get a little bit more complicating. Depending on the situation, the throughput can be increased with an increase in cores but there are other situations where there is no difference.

ps. In case you don't know the term, throughput is the number of tasks which are completed over a period of time. Usage of this term might better capture what you mean by "computing power", since you're probably worried about how much faster your games can run given the CPU.

Another thing to point out, is that with the usage of SMP technology, like in the multi-core processors, you can lose some performance that you would otherwise not have lost using a stand-alone processor. Like I said earlier, SMP has its applications. It depends on the processes are built. There are always physical limitations to how your programs execute on your computer. A processor can only run one process at a time, and running one process on multiple processors is difficult and requires multi-threading.
If the software which you are using doesn't implement threads or multiple processes working together on the same goal, then it won't make use of multiple processors. If you have a single processor running at 2.0 GHz, you're limited to just the one processor so only one program can execute within it at a time. Yet, with two processors running at 1.0 GHz, you're open to run two programs at once - just at half the speed.

Unless your program is really mulitple processes or mult-threaded, it can only run on one of the two 1.0 GHz processors. So, individually it'd run better with the 2.0 GHz processor. Yet, since you're not likely to run only just one program on a computer (which often times has many processes running at once in memory) you might experience a speed increase when using the program because it doesn't have to share the CPU time so much with other processes - having a spare processor around.

So, it's really hard to answer your question. I don't know if they provide statistical analysis, showing the differences between the two, but I do think that Intel has done studies which show that the duo-core processors out-perform the traditional single-core processors.
 
Last edited:
  • #3
How much can I overclock an AMD Phenom X4 2.2 GHz without needing some really powerful cooling system? Could I get 2.8, or 3.0 GHz out of it? Or at least maybe a few more hundred MHz?

Will the new CPU architecture, such as the Intel Core 2, and AMD Phenom, running at sub 3,000 MHz, with 8 GBs of RAM, run programs like EA's "Battlefield 2" much faster, smoother, and more efficiently than it would running on a Pentium 4, 3.0 GHz with 1 GB of RAM? Will I get a much better framerate and less video lag? Would Battlefield 2 run much better frames per second and video quality wise If I were running it on one of the latest DirectX 10 graphics cards? e.g. Radeon HD4870 X2, or GeForce GTX-280.

I heard that windows vista running in the background alone uses about 3+ GBs of RAM.

Are there any multi-core processors that use hyperthreading currently?

Would Crysis run pretty good on a 2.2 GHz Phenom X4 with 8GB of RAM? Or do I need to up the CPU clock speed to get decent performance and framerate out of the game? Doesn't Crysis actually utilize multiple cores?

When you compare quad-core AMD Phenoms against quad-core Intel Core 2's, which is faster overall? Especially for gaming?

What kind of graphics card(s) would I need to run Crysis on all "very high" detail settings? Will a single Radeon HD4870 X2 do the job? If I wanted to run it on higher resolutions would I need to add in more graphics cards in SLI?
 
  • #4
The_Absolute said:
Will the new CPU architecture, such as the Intel Core 2, and AMD Phenom, running at sub 3,000 MHz, with 8 GBs of RAM, run programs like EA's "Battlefield 2" much faster, smoother, and more efficiently than it would running on a Pentium 4, 3.0 GHz with 1 GB of RAM?

I'm not sure how the architecture (Pentium 4 vs. Phenom vs. Core 2) affects those games -- I'd think not much. The increase in RAM will help if the games + background services require more than 1 GB of memory, which seems likely. There's probably a reasonable performance increase when you jump from 1 GB to 2 GB, and a small jump when you go from 2 GB to 8 GB.

The_Absolute said:
I heard that windows vista running in the background alone uses about 3+ GBs of RAM.

Nonsense. 800 MB, tops. If you strip out the features for performance you can get that below 500 MB without much trouble -- I do that at work.

The_Absolute said:
Are there any multi-core processors that use hyperthreading currently?

Yes, but you wouldn't want them for gaming. For your purposes hyperthreading is probably counterproductive.

The_Absolute said:
Would Crysis run pretty good on a 2.2 GHz Phenom X4 with 8GB of RAM? Or do I need to up the CPU clock speed to get decent performance and framerate out of the game? Doesn't Crysis actually utilize multiple cores?

The processor matters much less than the GPU with Crysis and other similar games.

The_Absolute said:
When you compare quad-core AMD Phenoms against quad-core Intel Core 2's, which is faster overall? Especially for gaming?

The Core 2 wins that fight. Phenoms unfortunately haven't been performing particularly well, so I'd avoid them unless there's a good price difference. But the graphics cards matter more; either processor will do what you need.

The_Absolute said:
What kind of graphics card(s) would I need to run Crysis on all "very high" detail settings? Will a single Radeon HD4870 X2 do the job? If I wanted to run it on higher resolutions would I need to add in more graphics cards in SLI?

If you're running at "very high" and with high resolution, dual graphics cards would be useful. In any case if framerate is what you want, focus on the graphics cards not the processor -- get a cheaper processor if you need to save money for the graphics cards.
 
  • #5
First off: Intel vs AMD. If performance is all (aka if you aren't a fanboi type person) then Intel CPUs are currently faster.
Secondly: 3.3 vs 2.2/2.3 - While your old single core had a higher clock speed, the quads will be much faster, on both single and multithreaded apps. It's obvious why they'd be faster on multithreaded apps, but if you're confused as to why they'd be faster on single threaded apps (where they can only use 1 of their 4 cores) it is for multiple reasons: A clock speed isn't actually an accurate reading of performance, as it only represents the number of CPU cycles, not data throughput of each CPU cycle. Also another huge factor is microarchitecture (which has changed significantly since the days of the Pentium 4). And then there's transistor size: your P4 would be using 90nm transistors, as opposed to newer CPUs which are using 45 and 65 (Intel & AMD respectively) nanometre transistors. And let's not forget the Cache, which has gotten much larger since the P4.

To sum up, I'd go with the Intel machine based on performance (not taking price into consideration). And whichever one you go with, your new CPU will be much, much, much faster than your old one.
 
  • #6
The_Absolute said:
Will the new CPU architecture, such as the Intel Core 2, and AMD Phenom, running at sub 3,000 MHz, with 8 GBs of RAM, run programs like EA's "Battlefield 2" much faster, smoother, and more efficiently than it would running on a Pentium 4, 3.0 GHz with 1 GB of RAM? Will I get a much better framerate and less video lag? Would Battlefield 2 run much better frames per second and video quality wise If I were running it on one of the latest DirectX 10 graphics cards? e.g. Radeon HD4870 X2, or GeForce GTX-280.

I heard that windows vista running in the background alone uses about 3+ GBs of RAM.
Well, Vista is separate of the hardware issue at hand.
The_Absolute said:
Are there any multi-core processors that use hyperthreading currently?
I think that's kind of what the definition of "multi-core" implies. I know that Intel's hyperthreading technology provided "virtual processors" of which the OS had to be aware of. This way, a multithreaded process could have its threads different threads of execution at once. The reason why you couldn't apply this to actual processes (programs in execution) is that processes have their own unique data. You'd end up storing data which meant one thing to one program in the same place of another.

In spite of what another person on this thread said, you can have a single process use up all cores in a processor, but it'd likely just use all four of them as if it were a single CPU if it only had one thread of execution. Whereas a process which had maybe 8 threads of execution could use at least run four of its eight threads if the CPU had concessions to run four threads simultaneous, such as a CPU having two cores - each hyperthreaded for two threads on each core.

I think the biggest difference between hyperthreading - and this is is just a guess based on what I know - is that multi-core processors are a step above the single-core hyperthreaded cpu's, because it could support stand-alone processes. However, they'd probably share the same data and would differ little from threads (with the exception of having its own unique address space).

The_Absolute said:
How much can I overclock an AMD Phenom X4 2.2 GHz without needing some really powerful cooling system? Could I get 2.8, or 3.0 GHz out of it? Or at least maybe a few more hundred MHz?
I don't have a clue dude. That sounds more like an engineering/physics issue. I'm someone who's speaking from a computer science pov as a student of computer. But, I guess a person who's actually tried it before could tell you, but this question isn't a question for this room as it goes beyond the scope of computer science.

The_Absolute said:
Will the new CPU architecture, such as the Intel Core 2, and AMD Phenom, running at sub 3,000 MHz, with 8 GBs of RAM, run programs like EA's "Battlefield 2" much faster, smoother, and more efficiently than it would running on a Pentium 4, 3.0 GHz with 1 GB of RAM?
That's pretty hard to say. If the prior is "sub [3.0 GHz]" then it depends on how low below 3.0 GHz we're talking about. More memory is always better to have, because it means there's more room for your processes to sit there while they're waiting for the CPU and also less of a chance you have of your OS putting some of your process's data onto storage disk (which is likeliest one of the slowest forms of memory you'll have on your pc).
 
Last edited:
  • #7
Gr!dl0cK said:
A clock speed isn't actually an accurate reading of performance, as it only represents the number of CPU cycles, not data throughput of each CPU cycle. Also another huge factor is microarchitecture (which has changed significantly since the days of the Pentium 4). And then there's transistor size: your P4 would be using 90nm transistors, as opposed to newer CPUs which are using 45 and 65 (Intel & AMD respectively) nanometre transistors. And let's not forget the Cache, which has gotten much larger since the P4.

I don't know that the microarchitechture will be that much of an improvement, although by all accounts it has changed significantly. For example, one of the more important instructions for me (doing number theory) is IDIV, integer division and remainder. The Pentium 4 performs this operation with a throughput of 23 clock cycles, compared to 30 or 23 for Core 2 0F_3H and 0F_2H, respectively. That means (depending on your stepping) a Pentium 4 at 3 GHz can do 130 million divisions in a second, while a Core 2 0F_3H at 2.66 GHz can do only 89 million.

Glancing at the SIMD instructions (more likely to be used in games), the story is similar: the instructions I looked at were mostly the same between processors (both for throughput and latency), with an edge to the Pentium 4.


Having said that, you have a good point with the cache: the newer processors have much larger caches that will probably significantly benefit some applications. Are games cache-limited? Graphics surely isn't -- the amount of memory used there is too large to be effectively cached regardless of the L2+L3 size. But game logic, and possibly special effects like dust clouds or snow, may be too large for small caches but fit into large caches, so I could see a good improvement there.

The fab type isn't directly related to performance at all -- whatever advantage there is turns in in clock speed. And note that Intel certainly still uses the 65nm processes: I just bought a nice new server with 65nm processors.
 
  • #8
FTR: Hyperthreading is different from multithreading (you should read up on it).

PS: from what I've seen, microarchitecture does have a large impact on performance.

BTW [CRGreathouse]: Intel's more recent CPUs use 45nm, but they still make some (older ones) that use 65nm. Just as AMD's latest use 65, but they still make some (older ones) that use 90nm. And Core 2s (and the new i7) will have a much higher data throughput with each CPU cycle, giving them better bandwidth, despite the fact that each core may have less cycles.
 
  • #9
Gr!dl0cK said:
from what I've seen, microarchitecture does have a large impact on performance.

It does for me. I'm less sure (as a non-gamer) how much it affects gamers.

Gr!dl0cK said:
BTW [CRGreathouse]: Intel's more recent CPUs use 45nm, but they still make some (older ones) that use 65nm. Just as AMD's latest use 65, but they still make some (older ones) that use 90nm. And Core 2s (and the new i7) will have a much higher data throughput with each CPU cycle, giving them better bandwidth, despite the fact that each core may have less cycles.

Of course Intel makes 65nm chips -- that was my point. :-p

But I'm still not sure what you mean by the higher throughput of Core 2's. (i7 is a different story; no disagreement there.) I just quoted Intel manuals claims of throughputs for several operations' throughputs, and the Pentium 4 was as good or better in every case. Are you using throughput differently than I am, or is there an important class of instructions where the Core 2 throughput is better?
 
  • #10
Gr!dl0cK said:
FTR: Hyperthreading is different from multithreading (you should read up on it).

That came off to me as smug, as if I'm some idiot posting some bs and you're a know-it-all. Well, for your information, I have looked up on it already, as what I said is confirmed in my textbook. According to Silberschatz, Galvin, and Gagne, "SMP [Symmetric Multiprocessing] systems allow several threads to run concurrently by providing multiple physical processors. An alternative strategy is to provide multiple logical - rather than physical - processors. Such a strategy is known as symmetric multithreading (or SMT); it has also been termed hyperthreading technology on Intel processors."
 
Last edited:

Related to CPU comparisons (I need your help)

What is a CPU?

A CPU (Central Processing Unit) is the primary component of a computer that performs basic arithmetic, logical, control, and input/output operations specified by the instructions of a program.

Why are CPU comparisons important?

CPU comparisons are important because they help individuals make informed decisions when purchasing a computer or upgrading their current one. By comparing different CPUs, one can determine which processor is best suited for their needs and budget.

What factors should be considered when comparing CPUs?

The main factors to consider when comparing CPUs are clock speed, number of cores, cache size, and architecture. Clock speed determines how fast a CPU can process instructions, while the number of cores and cache size affect multitasking and overall performance. Architecture refers to the design and technology used in a CPU, which can also impact performance.

What is the difference between Intel and AMD CPUs?

Intel and AMD are two different companies that produce CPUs. The main difference between them is in their architecture and design. Intel CPUs are known for their high single-core performance, while AMD CPUs are known for their multi-threaded performance and affordability.

How do I know which CPU is best for my needs?

The best CPU for your needs depends on your specific requirements and budget. Consider the tasks you will be using your computer for and research the recommended CPU specifications for those tasks. It is also helpful to read reviews and benchmarks of different CPUs to get an idea of their performance.

Similar threads

  • Computing and Technology
Replies
2
Views
3K
  • Programming and Computer Science
Replies
9
Views
4K
  • Computing and Technology
Replies
6
Views
5K
Replies
9
Views
7K
  • Computing and Technology
Replies
7
Views
2K
  • Computing and Technology
Replies
2
Views
4K
Replies
1
Views
3K
  • Computing and Technology
Replies
13
Views
13K
  • Computing and Technology
Replies
2
Views
4K
  • Computing and Technology
Replies
1
Views
2K
Back
Top