The Truth: i5 750 vs Phenom II 965 -Updated

[AMW1011]

Hello
There has been a lot of heated debate and arguing on these forums about which is better, a Phenom II 965 or an i5 750. On both sides, everyone has failed to back up their point and have posted one or two links at best which is inadequate to answer such a question. I am here to answer that question. The scientific method tells us to do everything in, at least, threes. Three sources, three experiments, three variables, ect. Well I have compiled 8 different sources. None were picked and chosen except for the Lost Circuits article since it keeps getting thrown around.

I must remind everyone that this is a thread to inform those who are trying to choose between the two, I'm not here to settle any vendettas or disputes, I don't give a rats ass about that.

Alright let us discuss Turbo mode. Turbo Mode is where, with 1-2 cores under load, the CPU clocks those cores to 3.2 GHz and with 3-4 core under load the CPU clocks to 2.8 GHz. With Turbo mode enabled, which it is in all the below benchmarks, the i5 750 clocks at 2.8 GHz for most of the app benchmarks, which is 600 MHz (2.8GHz vs 3.4 GHz) lower than the Phenom II 965. In most games the i5 750 is clocked at 3.2 GHz or 2.8 GHz which is 200 MHz or 600 MHz (3.2GHz/2.8GHz vs 3.4 GHz) lower than the Phenom II 965. A reader must take this into account with the below results.

Links:
http://www.techspot.com/review/193-intel-core-i5-750/
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3634
http://www.guru3d.com/article/core-i5-750-core-i7-860-870-processor-review-test/
http://www.overclockersclub.com/reviews/intel_corei5750_corei7870/
http://www.neoseeker.com/Articles/Hardware/Reviews/lynnfield/
http://www.bit-tech.net/hardware/cpus/2009/09/08/intel-core-i5-and-i7-lynnfield-cpu-review/1
http://www.tomshardware.com/reviews/intel-core-i5,2410.html
http://www.lostcircuits.com/mambo//index.php?option=com_content&task=view&id=63&Itemid=42

Note: No synthetics were listed here, though I think that they have some merit, the majority opinion is that they do not and therefor they are not included.

Here we go: (continued on next post)

2/24/2010 I have updated the pricing section and some facts. If I'm asked I will add some more articles that are more recent.

 

[jennyh]

Didn't I just read on the page before that you were agreeing to the i5s superior performance and said you have thought that way all along?

Or am I confused? :/

No. I admire AMW's attempt but it has a lot of flaws. It doesn't compare motherboards etc - I bet that many of those game benchmarks were done on an inferior AMD mobo while almost all of them would be on top of the range intel mobo's.

That's just one example, the compiler is another.

It's a good list but it only validates what I already knew, that the 965 is a faster cpu, however slightly.

 

[Raidur]

At about 3.6 GHz both architectures become just as close and the i5 750 cannot be pointedly recommended because of overclocking.

That's almost a full ghz for i5 (not including turbo, which would make it an 800mhz difference on quad optimized) and only 200mhz for phenom ii. Is clock speed scaling that bad on i5 or do you mean in gaming (gpu bottlenecked situation)?

Id call twice as much oc headroom something to consider when picking but *shrug. For me (because I oc) it makes the decision easy.

Love the main post btw. Well done, I'm glad I didn't have to do it, but it needed to be done.

 

[Raidur]

No. I admire AMW's attempt but it has a lot of flaws. It doesn't compare motherboards etc - I bet that many of those game benchmarks were done on an inferior AMD mobo while almost all of them would be on top of the range intel mobo's.

That's just one example, the compiler is another.

It's a good list but it only validates what I already knew, that the 965 is a faster cpu, however slightly.

I was wondering when you were going to say that.

Maybe you should be the one to prove all of us (and paid reviewers...) wrong jenny.

 

[AMW1011]

Intel's compiler has been used for software other than synthetics. The point I'm making here is, 1 in 10 programs have been compiled with the intel compiler, making 1 in 10 programs artificially slower on AMD systems. With both cpu's being so close in your results, what do you think is the fastest when you take out that 10%?

As for 32nm, I haven't seen anything pointing to a 2010 launch of AMD 32nm quads either, maybe you can post something?

I don't care if Intel had an advantage by sprinkling fairy dust on their processors, if it makes it faster in real world apps, regardless of why, then IT IS FASTER. You can't just ignore being FASTER because you don't like why it is faster.

Thanks. 😍

No but I can point to Thuban coming soon and plugging straight into the same mobo.

No. I admire AMW's attempt but it has a lot of flaws. It doesn't compare motherboards etc - I bet that many of those game benchmarks were done on an inferior AMD mobo while almost all of them would be on top of the range intel mobo's.

That's just one example, the compiler is another.

It's a good list but it only validates what I already knew, that the 965 is a faster cpu, however slightly.

Yes, every single review has an inferior motherboard, lets see:

Techspot: Asus M3A79-T Deluxe decent board, I see no reason for any bottlenecks here
Anandtech: AMD 790FX again another one of the better AMD chipsets
OCC: Gigabyte MA790FXT-UD5P wow one the best AM3 boards
Neoseekers: Gigabyte MA790FXT-UD5P again
Bit-tech: Gigabyte GA-MA790FXT-ED5P again...
Tomshardware: Asus M4A79T Deluxe Another great board...

Too bad that theory didn't hold water, at all.

Turns out, the i5 750 is the faster CPU, however slightly...

 

[jennyh]

And how would you explain away the 10% of ALL software that has been compiled with the intel compiler?

 

[Raidur]

Jenny is impossible amw, you know this. You couldn't spend 12 hours and convince her.

 

[AMW1011]

While this is true, this thread is not about her and is about those who need help choosing. If someone spread misinformation, I will correct it for those who read it not for me or the one spreading the misinformation.

And Jenny, again, even if Intel has a shady way of making 10% of software faster on its processors, its still faster. Assuming that it shouldn't be does not change the fact that it is. IT.IS.FASTER. I'm not sure what you are trying to prove.

 

[jennyh]

Simple.

The FTC have demanded that intel fix that 10% of software which they bought illegally.

Btw - what happens if you remove all of Anandtech's and THG's results? Does the Phenom II suddenly become the faster cpu?

 

[JAYDEEJOHN]

Im letting the FTC to rule on this one, as theyll look into it in more depth than Im willing to, and if its the case, it could cost Intel some as well here, if they find it cheating

 

[randomizer]

And how would you explain away the 10% of ALL software that has been compiled with the intel compiler?

Where is this number from? I've been looking for hard numbers on this for ages and I always just get hearsay from 1% to 80%. I know the FTC mentioned software from certain companies but I don't remember a particular number. Assuming it is true, just because 10% of programs are compiled with the Intel compiler does not mean that 10% of the programs in this thread were compiled with the Intel compiler. The majority would have been MS Visual C++ and GCC/G++ (the latter for the open source programs). I am unsure whether POV-Ray binaries use GCC or ICC as it has been stated on the POV-Ray newsgroups that it runs about 10% faster when compiled with the Intel compiler, but no official documentation makes note of what is used. It would be reasonably safe to assume that GCC was used because ICC costs money, but to be absolutely sure the reviewers could always compile the source with both compilers themselves and compare the benchmark speeds with the binaries.

 

[yannifb]

Actually at 3.2 Ghz and up the Phenom II is pretty much equal to or better than an i7 when it comes to games. Scaling for the i7 is also pretty bad as a 4Ghz PhII is just as fast/faster than a 4Ghz i7 in games. Proof:
http://www.legionhardware.com/articles_pages/cpu_scaling_with_the_radeon_hd_5970,1.html

 

[jennyh]

Intel themselves said they were a 'minor player in the compiler market, owning only 10%'. It's actually the only time I've ever seen intel own up to not being #1 lol.

I'll try to find it.

 

[AMW1011]

Simple.

The FTC have demanded that intel fix that 10% of software which they bought illegally.

Btw - what happens if you remove all of Anandtech's and THG's results? Does the Phenom II suddenly become the faster cpu?

Anandtech's results were in-line with every other the review. The only one that didn't quite add up was the Tomshardware review, but only in 2 benchmarks.

The FTC can demand no such thing, even if they prove that Intel bought off anyone, which they haven't. It is the software company's prerogative to use whatever they choose and can not be penalized and forced to switch their technologies because of possible corruption form Intel. It can't happen. Sorry.

 

[jennyh]

Uhh, maybe you want to read the FTC's complaint against intel before deciding what 'can't happen' AMW.

Starting here

11-
Requiring Intel to correct the deceptive or misleading statements and omissions it has made in the past.

 

[Raidur]

"Perhaps a more important comparison is between the Core i5 750 and the Phenom II X4 processors. Again when clocked at either 2.80GHz or 3.60GHz the Core i5 750 smashed the Phenom II X4 965."

Your very own legion hardware saying it jen.

http://www.legionhardware.com/articles_pages/clock_for_clock_core_i5core_i7core_2_quad_and_phenom_ii_x4_performance,13.html

I know this is clock for clock but since they oc to about the same this is relevant to most of us.

 

[randomizer]

Intel themselves said they were a 'minor player in the compiler market, owning only 10%'. It's actually the only time I've ever seen intel own up to not being #1 lol.

I'll try to find it.

Given the vast amount of software in existence, 10% seems like a pretty exaggerated number. It depends if they mean 10% of all unique programs or 10% of the total software install base. In the corporate/HPC market you may have thousands of copies of the same program installed by just one company. They may not be talking about consumer software, in which case 10% of non-consumer software is quite a sound figure.

 

[Raidur]

Actually at 3.2 Ghz and up the Phenom II is pretty much equal to or better than an i7 when it comes to games. Scaling for the i7 is also pretty bad as a 4Ghz PhII is just as fast/faster than a 4Ghz i7 in games. Proof:
http://www.legionhardware.com/articles_pages/cpu_scaling_with_the_radeon_hd_5970,1.html

That means that it takes 3.2ghz to match i7 at that resolution with that card. This is irrelevant to anyone using anything under a 30" 26x16 resolution, which is most of us. Lower resolutions ( that most of us use ) might tell a different story than that review, however none of us have found a reliable one with a 5970.

It also does not mean scaling is bad for the i7, it means that the i7 needs less than phenom ii to run the 5970 to the max. Yes phenom ii does sometimes win when the gpu bottleneck is reached but its never more than a few frames.

 

[AMW1011]

Uhh, maybe you want to read the FTC's complaint against intel before deciding what 'can't happen' AMW.

Starting here

11-
Requiring Intel to correct the deceptive or misleading statements and omissions it has made in the past.

11-
Requiring Intel to correct the deceptive or misleading statements and omissions it has made in the past.

12 -
Prohibiting Intel from coercing or influencing benchmarking organizations to adopt benchmarks that are deceptive or misleading.

Intel must admit their mistake, developers must do nothing. And all that is if Intel is found guilty according to this article.

To remedy the anticompetitive damage alleged in the complaint, the FTC is seeking an order which includes provisions that would prevent Intel from using threats, bundled prices, or other offers to encourage exclusive deals, hamper competition, or unfairly manipulate the prices of its CPU or GPU chips. The FTC also may seek an order prohibiting Intel from unreasonably excluding or inhibiting the sale of competitive CPUs or GPUs, and prohibiting Intel from making or distributing products that impair the performance–or apparent performance–of non-Intel CPUs or GPUs.

NOTHING about developers changing a damn thing, and nothing about Intel changing the pre-existing implementation, only to stop any supposed acts of unfair market manipulation. They are seeking to stop Intel from buying developers, not distributing the compiler. The complier isn't even a large point in it, its just a single paragraph:

In addition, allegedly, Intel secretly redesigned key software, known as a compiler, in a way that deliberately stunted the performance of competitors’ CPU chips. Intel told its customers and the public that software performed better on Intel CPUs than on competitors’ CPUs, but the company deceived them by failing to disclose that these differences were due largely or entirely to Intel’s compiler design.

That is it. Notice that even they say that they have alleged reports and accusations. They only mention this once, and in passing. They say it with the term allegedly, meaning they have no evidence. Furthermore, they don't seem to be all that worried about the impact since it is only a single paragraph with no action or penalties implied or stated.

Start here:
http://www.ftc.gov/opa/2009/12/intel.shtm

Now, back to your article.

This imposes a severe performance handicap for all non-Intel CPUs. This would not be an issue if Intel's compilers were rarely used, but Intel's compilers very often produce the fastest binaries, particularly on Intel parts, so many performance sensitive applications use them." There is no doubt that Intel's investment in developing its own compiler and compiler tools gives them the benefit of hand-tuning the compiler code to run better on their own products. After all, AMD's GCC Linux compilers do the same thing for their own processors.

Seems that the FTC don't believe that the compiler is used because Intel pays off devs, but because it it very often produces the fastest binaries, even on non-Intel parts.

Sorry, this whole compiler BS holds no water.

 

[jennyh]

Sure that's why the FTC are using it in their complaint.

Seriously now. The reason this is going to court is because the FTC have told intel to make right what they cheated AMD and the consumer out of.

It doesn't matter if intel's compiler is fastest, all that matters is that it gives an unfair advantage to intel when used.

Do you use firefox? How would you feel if it ran 50% slower just because it was compiled on microsoft's compiler? That is what we are talking about here - the unfair crippling of competition because they have no fair answer to it.

The i5 is a slower cpu AMW, and your numbers prove it to anybody who knows what is really going on. I would have argued it before, but frankly I'm sick of i5 vs phenom II.

http://news.cnet.com/8301-1001_3-10417256-92.html

# Intel can't make hardware or software designed to inhibit processors made by competing companies.
# For customers who bought "defective" compilers, Intel must provide them with a working compiler at no cost and compensate them for the cost of recompiling their software using the new compiler.
# Intel can't coerce benchmarking organizations to adopt benchmarks that are deceptive or misleading.

 

[keithlm]

The majority would have been MS Visual C++ and GCC/G++ (the latter for the open source programs).

You guys do realize that the popular Intel libraries such as IPP, MKL, VML and SVML can be used with the Microsoft Visual C++ and GCC/G++ compilers? (Or visual basic, or even Java?)

I'll give you one guess what happens if you choose to use these libraries... and it doesn't matter WHICH compiler you use. Yep... GenuineIntel will have the benefit of the CPU extensions and non-GenuineIntel will always use the slowest possible extension.


http://software.intel.com/en-us/articles/using-intel-ipp-with-different-programming-languages/

So the number of 10% that has been thrown around is probably way too conservative.

 

[randomizer]

You guys do realize that the popular Intel libraries such as IPP, MKL, VML and SVML can be used with the Microsoft Visual C++ and GCC/G++ compilers? (Or visual basic, or even Java?)

I am aware of that, but we don't know what software uses these libraries so drawing conclusions that everything is neutered on AMD processors is a pretty big stretch. Also, since you have to pay for all these libraries and they are not open source (as far as I know), FOSS developers are unlikely to use them. I would not put it past commercial software developers using MS Visual C++ or other compilers though.

 

[keithlm]

so drawing conclusions that everything is neutered on AMD processors is a pretty big stretch.

Claiming that everything is neutered on AMD processors would be showing just as much ignorance as claiming that nothing is neutered and that all of this is a non-issue.

 

[AMW1011]

Sure that's why the FTC are using it in their complaint.

Seriously now. The reason this is going to court is because the FTC have told intel to make right what they cheated AMD and the consumer out of.

It doesn't matter if intel's compiler is fastest, all that matters is that it gives an unfair advantage to intel when used.

Do you use firefox? How would you feel if it ran 50% slower just because it was compiled on microsoft's compiler? That is what we are talking about here - the unfair crippling of competition because they have no fair answer to it.

The i5 is a slower cpu AMW, and your numbers prove it to anybody who knows what is really going on. I would have argued it before, but frankly I'm sick of i5 vs phenom II.

http://news.cnet.com/8301-1001_3-10417256-92.html

# Intel can't make hardware or software designed to inhibit processors made by competing companies.
# For customers who bought "defective" compilers, Intel must provide them with a working compiler at no cost and compensate them for the cost of recompiling their software using the new compiler.
# Intel can't coerce benchmarking organizations to adopt benchmarks that are deceptive or misleading.

Jenny this is getting ridiculous. We are debating an FTC case so you link a Cnet article to prove my source wrong, the actual god damn FTC website. Now your making up more numbers. Just give it a rest, you love AMD and hate everything Intel we get it. Now let us continue with reality and we will let you continue with your fiction.

 

[jennyh]

I am aware of that, but we don't know what software uses these libraries so drawing conclusions that everything is neutered on AMD processors is a pretty big stretch.

That is why I used the 10% market share that intel said they had.

When you take two cpu's closely matched like the i5 750 and 965 BE, it becomes clear just how much that "10%" is really worth - if it can turn a losing cpu into a 'winning' one - because that is exactly what the i5 is compared to a 965.

 

[jennyh]

Jenny this is getting ridiculous. We are debating an FTC case so you link a Cnet article to prove my source wrong, the actual god damn FTC website. Now your making up more numbers. Just give it a rest, you love AMD and hate everything Intel we get it. Now let us continue with reality and we will let you continue with your fiction.

You need to get a better understanding of the situation. Not gonna say any more AMW, I have no problem with you and I wouldn't want your lack of understanding of the subject to change that.