Intel Core i5 750, Core i7 870 Linux Benchmarks

[phoronix]

Phoronix: Intel Core i5 750, Core i7 870 Linux Benchmarks

Now that we have provided a brief overview of the Intel P55 and how it functions under Linux, our larger area of concentration is looking at the Linux performance of the P55 with the new Core i5 750 and Core i7 870 processors. We have a number of benchmarks in this article along with more information on these Lynnfield processors.

http://www.phoronix.com/vr.php?view=14174

[Dark_Star]

Awesome :) Thanks for the review..

So it seems my Phenom II X4 will have significant lead over these i5 and i7 :)

Could you please include some Phenom II X4 955 benchmarks.

[vermaden]

What sense in comparing Intel quad core CPU to AMD tri core CPU?

You should at least include Intel Core 2 Quad CPU (to compare architecture improovements) and AMD Phenom II X4 (to compare new Intel CPUs to REAL competiotion, not to some weaker and cheaper tripple core Phenom II X3).

These whole benchmarks just compare Intel i5 to Intel i7 CPUs performance and nothimg more ... really disapoining.

Why waste time at all for such uncomplete tests?

[Pepazdepa]

These whole benchmarks just compare Intel i5 to Intel i7 CPUs performance and nothimg more ... really disapoining.

Why waste time at all for such uncomplete tests?

From my point of view, I can se Phenom X3 710 kicking Intel into his core i5 butt:D

[Dark_Star]

I have ask amd senior executive to send some CPUs to Phoronix. But I would suggest you to ask your hardware vendor to lend them for testing.

A Quad should be tested with Quad. :)

[Kano]

The benchmarks are interesting, where is the pts code to compare against? Also adding Phenom II X965, Intel Q9550, would be interesting, as they are from the same price range, i7 860 instead of 870 would be more logical too, as 870 is much more expensive than i7 920/860.

[lem79]

Dear God, putting AMD's tricore against Intel's newest quads? Poor AMD. Michael you should hit them up for a 955 at least! Very surprised to see the X3 outperforming the Nehalems in places.

[lordmozilla]

Nice review, I'm guessing you only had that tricore at hand. But it really did do well. I was very impressed with its performance compared to Intel's i5 and i7.

Any chance you could add a Top of the range Phenom II X4 in there?

[suokko]

Looks like phenom is very fast core but lags in memory bandwidth so test that are memory limited are lose to AMD processors.

[Hohlraum]

I think the only thing these benchmarks show is that Stream is the only software that accurately represents the performance of these processors over the AMD. I don't dislike AMD but I find it highly unlikely that it out performs ANY i7 processor on the market. If that is the case then Phoronix is the only review site I've ever seen that has come to this conclusion. :confused:

[Kjella]

Conclusion, new core i5/i7 work terribly under Linux. Least that's the only conclusion I can draw compared to everything I read at Anandtech.

Here's their main conclusion:
"I'll start this conclusion with what AMD must do in response to Lynnfield. The Core i5 750 is a great processor at $196, in fact, it's the best quad-core CPU you can buy at that price today. In nearly every case it's faster than AMD's Phenom II X4 965 BE, despite the AMD processor costing almost another $50." (...) "The Core i7 870 gets close enough to the Core i7 975 that I'm having a hard time justifying the LGA-1366 platform at all."

In short, the results I'm seeing here do not make sense. There's not a single case of the many, many Windows benchmarks where the i5 would lose to an triple-core processor.

[Apopas]

These whole benchmarks just compare Intel i5 to Intel i7 CPUs performance and nothimg more
That's the title of the benchmark after all. i5 750 vs i7 870 and not Intel vs AMD.

[lem79]

Phoronix is probably also the only site benchmarking on Linux? And using GCC instead of the Intel C compiler? I don't know if ICC still does dodgey stuff with checking CPU vendor strings etc, or if the version of GCC used doesn't properly support Nehalem ... but the results here should be a genuine representation of what they are. I doubt anyone went to any length to rig these results.

[Dragoran]

What is important to keep in mind though is that Intel Turbo Boost Technology was disabled on the processors during testing, since this functionality had not worked under Linux for increasing the clock frequency but instead appeared to cause some sporadic performance problems.

Err.. what?

Turbo mode does not really depend on the OS, and works fine on my i7 920.
And of course without turbo mode the single threaded performance is not quiet as good as it supposed to be, also comparing apples with oranges (quad vs. triple core) does not make the test very useful.

[suokko]

Phoronix is probably also the only site benchmarking on Linux? And using GCC instead of the Intel C compiler? I don't know if ICC still does dodgey stuff with checking CPU vendor strings etc, or if the version of GCC used doesn't properly support Nehalem ... but the results here should be a genuine representation of what they are. I doubt anyone went to any length to rig these results.

I wouldn't be surprised if icc is still doing some Intel only optimizations. It would be interesting to see difference wiht icc compiled tests to gcc compiled with all icc optimizations enabled. (I think some of optimizations would require some compiler switches.)

gcc has many contributions from amd while Intel only recently started contributing to gcc.

[Apopas]

Conclusion, new core i5/i7 work terribly under Linux.
Or maybe, Phenom work terribly under Windows :D

[Michael]

The amd tricore was used since that is just what was available...

[Max Spain]

If you look at some of the benchmarks, AMD has tended to be superior in anything involving cryptography, but I am still surprised to see a lowly AMD X3 pwn the bejeezus out of Intel's latest and greatest so frequently :eek:

[Ant P.]

Phoronix is probably also the only site benchmarking on Linux? And using GCC instead of the Intel C compiler? I don't know if ICC still does dodgey stuff with checking CPU vendor strings etc, or if the version of GCC used doesn't properly support Nehalem ... but the results here should be a genuine representation of what they are. I doubt anyone went to any length to rig these results.

If the i5/i7 are getting beaten by a Phenom II just because it's GCC, then maybe Intel should start contributing to GCC... considering it's what 99.9% of Linux users' software is built with.

[StringCheesian]

There's more info on page 4 of the motherboard review:On some test runs with the Phoronix Test Suite the Lynnfield processors would run very well, but when running the tests again just seconds later, the performance results would be wildly and severely impaired.

[...] core frequencies never increased to their Intel Turbo Boost frequencies when they were encountering a load. Intel's Turbo Boost Technology was not working under Linux. Once we disabled Turbo Boost from the BIOS, our sporadic performance problems were eliminated too. The performance numbers stopped fluctuating and dropping so much between runs and there were finally stable performance figures. Turbo Boost never boosted the performance under Linux or even the frequencies for that matter, but just seemed to cause some problems in our early testing.

Dragoran and other i7 owners, can any of you verify that your i7 is overclocking one or two cores when it should while running linux?

[mgc8]

There is something deeply wrong with the results of this benchmark! While the apparent wins of AMD processors can be attributed to anything from poor compiler support to the wild difference in platforms, there is no explanation for the unnatural balance of the Intel processors among themselves!

While the i7 920 coming out on top in so many benchmarks starts to feel fishy, seeing the i5 750 beat any of the other Intels is clearly wrong, and casts doubt on the whole test! I think the performance problems hinted in the beginning are definitely not solved by turning Turbo Mode off in the BIOS, and further investigation is necessary.

Perhaps these lines from the dmesg in the previous P55 article (http://www.phoronix.com/scan.php?page=article&item=intel_p55&num=3) give an indication:


ACPI Warning: \_PR_.CPU0._PSS: Return type mismatch - found Integer, expected Package 20090521 nspredef-940
[ 0.781763] ACPI: Invalid _PSS data
[ 0.781855] [Firmware Bug]: BIOS needs update for CPU frequency support
[ 0.781897] ACPI Warning: \_PR_.CPU2._PSS: Return type mismatch - found Integer, expected Package 20090521 nspredef-940
[ 0.781900] ACPI: Invalid _PSS data
[ 0.782000] [Firmware Bug]: BIOS needs update for CPU frequency support
[ 0.782040] ACPI Warning: \_PR_.CPU4._PSS: Return type mismatch - found Integer, expected Package 20090521 nspredef-940
[ 0.782044] ACPI: Invalid _PSS data
[ 0.782140] [Firmware Bug]: BIOS needs update for CPU frequency support
[ 0.782180] ACPI Warning: \_PR_.CPU6._PSS: Return type mismatch - found Integer, expected Package 20090521 nspredef-940
[ 0.782184] ACPI: Invalid _PSS data
[ 0.782283] [Firmware Bug]: BIOS needs update for CPU frequency support


There is also a factual error in the article: the i7 870 is most definitely not a "mid-range" processor, it is running at 2.93Ghz not 2.66 (that's the 750) and has a price of ~550$, not 250$ -- that's the i7 860. There is even an Amazon link that lists the price at 695$ (obviously inflated to get early-adopters). At that price, it would be foolish for Intel to try and sell the 870 if the 920 out-performed it, now wouldn't it?! Of course, every other benchmark everywhere places the 870 far ahead of the 920, it's obvious even from the frequency difference (2.93 vs. 2.66) that the 870 should do better! As it stands, the results of this particular benchmark can not be trusted -- except as indication of something very wrong with the new Core iX processors in Linux, which warrants a closer look.

Oh, and why is there no mention in either the P55 or this article about the other important change on the Lynnfields, integrating the PCI-E controller with 16x lanes on the processor die? That's the reason the QPI link was replaced by the slower DMI, and in itself could pose tricky problems for drivers and kernel developers, again an avenue of inquiry for the strange results of the tests.

Regards,
Mihnea

[energyman]

X3 710: 88€
i5 750: 172€

Valid conclusion would be:

the X3 710 is the price/performance winner, beating Intel's offerings in some tests in humiliating ways.

Hohlraum, there are many tests where AMDs Phenom 2 outperform Intel's i7:
http://www.tecchannel.de/_misc/galleries/detail.cfm?pk=30389&fk=380091&resize=no

[energyman]

well - maybe intel's offerigns just suck on linux?

looking at the bsf threads on lkml looks like there is something wrong in intel land.

So - lets say the results are valid. If you are a linux user, you shall buy AMD.

Who cares if Intel is faster in Windows, if they suck with the best of all operating systems?

[Ant P.]

Well to be fair, Intel is 100% dependent on Windows to sell CPUs so it's no surprise they'd build their CPUs around making it as fast as possible there.

They don't own the server market - the Pentium/Core FSB bottleneck has always kept it behind in multithreaded performance. Atom can't compete with ARM at all - and both Intel and MS know that. Intel is at least making an effort to keep customers with Moblin, whereas MS is (as usual) resorting to desperation tactics, paying off the likes of Asus and Bestbuy to spread FUD.

[Apopas]

Can someone give me a responsible answer about "with what compiler are the most commercial and free programs compiled under windows?"

[deanjo]

Can someone give me a responsible answer about "with what compiler are the most commercial and free programs compiled under windows?"

That would have to be the MSVS series one would have to assume.

[Apopas]

That would have to be the MSVS series one would have to assume.
That's what I thought as well. So is not Intel's compilers among the most popular in the windows world?

[sabby]

Just to add my 2 cents, I had mix feeling to see most review site post result with Turbo boost enabled. I personally would prefer to also see the result without. Nonetheless, I am quite surprised at the result posted on phoronix. Definitely will postpone buying one until at least things are sorted out, already do enough beta testing as is. Hopefully won't see a repeat of the problems with poulsbo...

[BillBroadley]

That would have to be the MSVS series one would have to assume.

It's worth nothing that Microsoft licensed the intel compiler technology to enable more aggressive optimizations.

To achieve similar performance numbers under linux you need to run Pathscale, Portland group, or the free open64 compilers to get similar numbers. For instance using said compilers gets over double the performance with stream when compared to the phoronix posted numbers.

[Apopas]

It's worth nothing that Microsoft licensed the intel compiler technology to enable more aggressive optimizations.

To achieve similar performance numbers under linux you need to run Pathscale, Portland group, or the free open64 compilers to get similar numbers. For instance using said compilers gets over double the performance with stream when compared to the phoronix posted numbers.
Is possible then, the Intel's proccessors to run better the programs that have been compiled with their own compiler?

[kotsergy]

One thing about comparison tables, it would've made more sense to arrange CPU results in order of increasing price from left to right for example, so that it is more practical and comfortable to compare the results. X3 then I7 then I5 then I7 does not make much sense.

[AdrenalineJunky]

when i saw they were comparing an X3 i rolled my eyes...

when i saw the results i nearly fell out of my chair in amazement.

WHAT? how does that even happen? in windows the i5 was obliterating the phenom 965....

[BillBroadley]

when i saw they were comparing an X3 i rolled my eyes...

when i saw the results i nearly fell out of my chair in amazement.

WHAT? how does that even happen? in windows the i5 was obliterating the phenom 965....

My best guess is that the compilers used under windows are more aware of the special features of the newest chips. Things like the double wide SSE, increased number of micro-ops possible, and various other optimizations. That and the combination of turning off the auto overclocking (mentioned in some detail in the p55 article) combine to show AMD with a substantial advantage.

So while this advantage is real *today*, with problems with intel's turbo boost, and compiling binaries with the default gcc.

There are compilers aware of such features. Portland Group, Pathscale, and the free open64. As an example compare the i7-920 results at:
http://www.phoronix.com/scan.php?page=article&item=intel_lynnfield&num=13

With my results using open64 on a i7-920 without overclocking(running ubuntu):

$ head -8 /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 26
model name : Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz
stepping : 4
cpu MHz : 2672.704
cache size : 8192 KB

$ ./stream-open64
-------------------------------------------------------------
STREAM version $Revision: 5.9 $
-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 20000000, Offset = 0
Total memory required = 457.8 MB.
...
Function Rate (MB/s) Avg time Min time Max time
Copy: 22410.3334 0.0143 0.0143 0.0143
Scale: 22282.7187 0.0144 0.0144 0.0144
Add: 22511.9469 0.0230 0.0213 0.0234
Triad: 20943.1595 0.0233 0.0229 0.0234

For comparison a Phenom II X4 810 (2.6 GHz) + DDR3-1333 using the same binary:
Function Rate (MB/s) Avg time Min time Max time
Copy: 12455.2457 0.0258 0.0257 0.0258
Scale: 12369.4995 0.0259 0.0259 0.0260
Add: 12539.4940 0.0384 0.0383 0.0387
Triad: 12442.0063 0.0387 0.0386 0.0387

Using gcc-4.4.1 on i7-920:
$ ./stream-gcc-4.4.1
Function Rate (MB/s) Avg time Min time Max time
Copy: 14374.3618 0.0223 0.0223 0.0224
Scale: 14416.3573 0.0222 0.0222 0.0223
Add: 15624.5172 0.0308 0.0307 0.0308
Triad: 15801.4749 0.0304 0.0304 0.0

Same on X4 810:
Function Rate (MB/s) Avg time Min time Max time
Copy: 8490.9773 0.0378 0.0377 0.0380
Scale: 8485.1263 0.0379 0.0377 0.0383
Add: 9569.1637 0.0503 0.0502 0.0508
Triad: 9573.1679 0.0505 0.0501 0.0528

For fun I'll run just 3 copies on the X4-810 to simulate using an X3:
Function Rate (MB/s) Avg time Min time Max time
Copy: 8326.3684 0.0386 0.0384 0.0389
Scale: 8329.6239 0.0386 0.0384 0.0389
Add: 9358.3690 0.0514 0.0513 0.0517
Triad: 9346.5083 0.0514 0.0514 0.0518

So as you can see using current compilers (default under windows or optionally under linux) can yield large differences in performance, in this case over 2x for the i7-920.

[deanjo]

My best guess is that the compilers used under windows are more aware of the special features of the newest chips. Things like the double wide SSE, increased number of micro-ops possible, and various other optimizations. That and the combination of turning off the auto overclocking (mentioned in some detail in the p55 article) combine to show AMD with a substantial advantage.

So while this advantage is real *today*, with problems with intel's turbo boost, and compiling binaries with the default gcc.

There are compilers aware of such features. Portland Group, Pathscale, and the free open64. As an example compare the i7-920 results at:
http://www.phoronix.com/scan.php?page=article&item=intel_lynnfield&num=13

With my results using open64 on a i7-920 without overclocking(running ubuntu):

$ head -8 /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 26
model name : Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz
stepping : 4
cpu MHz : 2672.704
cache size : 8192 KB

$ ./stream-open64
-------------------------------------------------------------
STREAM version $Revision: 5.9 $
-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 20000000, Offset = 0
Total memory required = 457.8 MB.
...
Function Rate (MB/s) Avg time Min time Max time
Copy: 22410.3334 0.0143 0.0143 0.0143
Scale: 22282.7187 0.0144 0.0144 0.0144
Add: 22511.9469 0.0230 0.0213 0.0234
Triad: 20943.1595 0.0233 0.0229 0.0234

For comparison a Phenom II X4 810 (2.6 GHz) + DDR3-1333 using the same binary:
Function Rate (MB/s) Avg time Min time Max time
Copy: 12455.2457 0.0258 0.0257 0.0258
Scale: 12369.4995 0.0259 0.0259 0.0260
Add: 12539.4940 0.0384 0.0383 0.0387
Triad: 12442.0063 0.0387 0.0386 0.0387

Using gcc-4.4.1 on i7-920:
$ ./stream-gcc-4.4.1
Function Rate (MB/s) Avg time Min time Max time
Copy: 14374.3618 0.0223 0.0223 0.0224
Scale: 14416.3573 0.0222 0.0222 0.0223
Add: 15624.5172 0.0308 0.0307 0.0308
Triad: 15801.4749 0.0304 0.0304 0.0

Same on X4 810:
Function Rate (MB/s) Avg time Min time Max time
Copy: 8490.9773 0.0378 0.0377 0.0380
Scale: 8485.1263 0.0379 0.0377 0.0383
Add: 9569.1637 0.0503 0.0502 0.0508
Triad: 9573.1679 0.0505 0.0501 0.0528

For fun I'll run just 3 copies on the X4-810 to simulate using an X3:
Function Rate (MB/s) Avg time Min time Max time
Copy: 8326.3684 0.0386 0.0384 0.0389
Scale: 8329.6239 0.0386 0.0384 0.0389
Add: 9358.3690 0.0514 0.0513 0.0517
Triad: 9346.5083 0.0514 0.0514 0.0518

So as you can see using current compilers (default under windows or optionally under linux) can yield large differences in performance, in this case over 2x for the i7-920.

Should be noted that Pathscale maybe no longer available in the near future. It all depends what happens now that Cray has bought it.

[Kano]

The i7 750 can be faster than the i7 920 because the i5 can use 5 levels of turbo boost and the i7 only 3. Each step is 133 mhz, some are only available with 1 or 2 cores. A benchmark like povray which only runs on 1 core should show it.

[AdrenalineJunky]

The i7 750 can be faster than the i7 920 because the i5 can use 5 levels of turbo boost and the i7 only 3. Each step is 133 mhz, some are only available with 1 or 2 cores. A benchmark like povray which only runs on 1 core should show it.

except that in this test turbo boost was disabled.

[justapost]

With my results using open64 on a i7-920 without overclocking(running ubuntu):

$ head -8 /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 26
model name : Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz
stepping : 4
cpu MHz : 2672.704
cache size : 8192 KB

$ ./stream-open64
-------------------------------------------------------------
STREAM version $Revision: 5.9 $
-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 20000000, Offset = 0
Total memory required = 457.8 MB.
...
Function Rate (MB/s) Avg time Min time Max time
Copy: 22410.3334 0.0143 0.0143 0.0143
Scale: 22282.7187 0.0144 0.0144 0.0144
Add: 22511.9469 0.0230 0.0213 0.0234
Triad: 20943.1595 0.0233 0.0229 0.0234

Those are triple channel results right? Max possible with dual channel ddr3 1333 would be 21200MB/s. I'd be interested in dual channel results, I expect something around 17GB/s. On an amd setup you must run the nb at around 2.6GHz and the cpu abit faster 2.8GHz to hit the max which was around 15.7GB/s here with DDR3 1333 CL7 (stream build with icc 11.1 and openmp support).

Update: Looking at your results it seemed to me open64 compilers generate as efficient code as icc 11.1 so I grabed the actual build and ran a comparison here.

PII 955BE 3.2GHz NB 2GHz MEM 2xDDR1333 Unganged CL7

ICC 11.1

Function Rate (MB/s) Avg time Min time Max time
Copy: 13223.4215 0.0026 0.0024 0.0150
Scale: 13261.3109 0.0025 0.0024 0.0090
Add: 13726.5011 0.0036 0.0035 0.0048
Triad: 13788.5482 0.0036 0.0035 0.0050


Open64 4.2.1

Copy: 8859.2560 0.0036 0.0036 0.0037
Scale: 8712.0426 0.0037 0.0037 0.0037
Add: 9541.0925 0.0050 0.0050 0.0051
Triad: 9749.9439 0.0056 0.0049 0.0111


GCC-4.3.1

Copy: 8820.2489 0.0041 0.0036 0.0055
Scale: 8544.5460 0.0041 0.0037 0.0054
Add: 9597.9497 0.0053 0.0050 0.0055
Triad: 9632.8513 0.0059 0.0050 0.0100


PII 955BE 3.2GHz NB 2.6GHz MEM 2xDDR1333 Unganged CL7


ICC 11.1

Function Rate (MB/s) Avg time Min time Max time
Copy: 15116.3113 0.0023 0.0021 0.0089
Scale: 15794.0372 0.0021 0.0020 0.0029
Add: 15185.2913 0.0032 0.0032 0.0035
Triad: 15320.4925 0.0032 0.0031 0.0050


Open64 4.2.1

Function Rate (MB/s) Avg time Min time Max time
Copy: 9624.1021 0.0054 0.0033 0.0216
Scale: 9329.0977 0.0035 0.0034 0.0035
Add: 10278.5823 0.0047 0.0047 0.0047
Triad: 10478.1197 0.0046 0.0046 0.0047


GCC-4.3.1

Function Rate (MB/s) Avg time Min time Max time
Copy: 9445.3011 0.0034 0.0034 0.0034
Scale: 9297.4320 0.0035 0.0034 0.0035
Add: 10225.8529 0.0047 0.0047 0.0048
Triad: 10405.5505 0.0046 0.0046 0.0047


I used openCC -fopenmp -O2 -o stream-o64 stream.c to build the open64 version. Seems I'm doing something wrong here.

[BillBroadley]

Those are triple channel results right?


Yes, triple channel.


Max possible with dual channel ddr3 1333 would be 21200MB/s. I'd be interested in dual channel results, I expect something around 17GB/s.


Alas my machine is in production, so I can't easily try dual channel. I've seen dual vs tri channel stream numbers posted in a hardware review, alas I can't remember where.


Update: Looking at your results it seemed to me open64 compilers generate as efficient code as icc 11.1 so I grabed the actual build and ran a comparison here.

PII 955BE 3.2GHz NB 2GHz MEM 2xDDR1333 Unganged CL7

ICC 11.1

Function Rate (MB/s) Avg time Min time Max time
Copy: 13223.4215 0.0026 0.0024 0.0150
Scale: 13261.3109 0.0025 0.0024 0.0090
Add: 13726.5011 0.0036 0.0035 0.0048
Triad: 13788.5482 0.0036 0.0035 0.0050


Open64 4.2.1

Copy: 8859.2560 0.0036 0.0036 0.0037
Scale: 8712.0426 0.0037 0.0037 0.0037
Add: 9541.0925 0.0050 0.0050 0.0051
Triad: 9749.9439 0.0056 0.0049 0.0111


I used openCC -fopenmp -O2 -o stream-o64 stream.c to build the open64 version. Seems I'm doing something wrong here.

Hrm, try this:

gcc -O4 -fopenmp stream.c -o s-gcc-4.3.3 -static
export PATH=/opt/pkg/gcc-4.4.1/bin:$PATH
gcc -O4 -fopenmp stream.c -o s-gcc-4.4.1 -static
export PATH=/opt/pkg/x86_open64-4.2.2.1/bin:$PATH
opencc -O4 -fopenmp stream.c -o s-open64-4.2.2.1 -static


Hopefully it will produce numbers like:

$ ./s-gcc-4.3.3 | grep Copy:
Copy: 8500.2266 0.0377 0.0376 0.0377
$ ./s-gcc-4.4.1 | grep Copy:
Copy: 8492.3205 0.0377 0.0377 0.0378
$ ./s-open64-4.2.2.1 | grep Copy:
Copy: 12487.2286 0.0258 0.0256 0.0258

[djiezes]

What is important to keep in mind though is that Intel Turbo Boost Technology was disabled on the processors during testing, since this functionality had not worked under Linux for increasing the clock frequency but instead appeared to cause some sporadic performance problems.

Update: after starting to see a flow of Windows-based reviews today, it looks like there are some more serious Linux + Lynnfield problems at hand, which we are currently investigating.

I thought the TurboBoost feature was OS-independent. It's a mainboard/BIOS setting isn't it? Why would linux not be able to handle this, while windows can?
Is this only a problem for the new Lynnfield (i5 750, i7 860 & 870) series, or do the core i7-9xx cpu's also share the same issues? I'm asking since the Turboboost feature underwent some changes between these cpus.

And what are the other linux+lynnfield problems you mention? Is it just the lm_sensors package that can't read the temps, or are there other problems at hand here?

Like many others, I planned on buying a Lynnfield core i7-860 soon, but if even an AMD triple core performs better in many tests for more than half the money, I'd want answers to above questions before making that decision.

Thanks for the benchmarks, but now i'm a bit disappointed with these indications for bad linux performance.

[justapost]

Yes, triple channel.

Alas my machine is in production, so I can't easily try dual channel. I've seen dual vs tri channel stream numbers posted in a hardware review, alas I can't remember where.
Thx for clarifying.



Hrm, try this:

opencc -O4 -fopenmp stream.c -o s-open64-4.2.2.1 -static


-O4 was the key, now I get those results wit the nb still at 2.6GHz.

Copy: 14486.5330 0.0025 0.0022 0.0031
Scale: 14246.6541 0.0026 0.0022 0.0046
Add: 14022.8872 0.0036 0.0034 0.0042
Triad: 14011.1763 0.0045 0.0034 0.0107

[movieman]

Atom can't compete with ARM at all - and both Intel and MS know that.

Atom can't compete in ARM's traditional low-power, low-cost, low-performance market, but ARM can't compete with Atom when running x86 code, which is still what most end-users want from a computer. ARM makes Atom look like a dinosaur in embedded systems, and the ARM netbooks look promising enough that I may buy one, but as soon as Joe Sixpack tries to install WoW on his ARM netbook and discovers it doesn't work, he'll be replacing it with an x86 of some description. Intel is still the 800lb gorilla in the CPU market and I can't see that changing any time soon.

I'd also add that building an ARM system for home use is insanely expensive compared to building an Atom system; I looked at ARM boards for a home server and the cost would have been at least twice as much as the Atom I eventually bought, for less capability and more hassle (e.g. needing a source of ARM Linux).

[gost80]

Could you also post some Java benchmarks? Using the sun java compiler. This I feel provides a fairly good estimate of performance as the binary is provided by sun and is always the same and I assume is very well optimized by them for multiple architectures.

Before I saw these benchmarks, I was totally kicking myself for ordering a P II X4 955 for my lab. This makes me feel much better.

[BillBroadley]

I thought the TurboBoost feature was OS-independent. It's a mainboard/BIOS setting isn't it? Why would linux not be able to handle this, while windows can?


The OS interacts with speedstep, you can do things like force a slow clock speed to extend battery life, lock it to full speed for benchmarking, and tune it in related ways based on your preferences. Intel's been pretty good with the documentation for GPU and CPU as of late, I suspect it's just an issue of a small amount of code or an updated table or two to fix.

Send a few lynnfields to a kernel maintainer or two with some docs and I'd expect it in a weekend.


Is this only a problem for the new Lynnfield (i5 750, i7 860 & 870) series, or do the core i7-9xx cpu's also share the same issues? I'm asking since the Turboboost feature underwent some changes between these cpus.


Dunno, I have a few nehalem around, but afaik turbo boost is off, for my needs I want consistent performance, not performance that depends on temp and/or what the other cores are doing. Hopefully someone else can contribute. I'd expect the nehalem review would mention this, I've not checked. I'd look for the phoronix nehalem review.


And what are the other linux+lynnfield problems you mention? Is it just the lm_sensors package that can't read the temps, or are there other problems at hand here?


All I know about linux and lynnfield is in the phoronix pair of articles on the p55 and lynnfield.


Like many others, I planned on buying a Lynnfield core i7-860 soon, but if even an AMD triple core performs better in many tests for more than half the money, I'd want answers to above questions before making that decision.

Thanks for the benchmarks, but now i'm a bit disappointed with these indications for bad linux performance.

If you are going to get an AMD you get a fair bit lower price in any case, probably save $80-$100 on the motherboard, and at least $50 (or in the case of the 860 $150 or so). Of particular interest to me is the new 95 watt 3.0 GHz phenom II for $170. I've got the same CPU at 2.6 GHz and it runs cool, fast, and is a pleasure to use. The phenom is especially nice if you want to use the built in graphics with the 785g, not a gaming machine, but very nice otherwise. Mine runs impressively cool, around 50 watts idle, and 110 watts under load, 115 or so if I really abuse it. Said config runs around $500 with 4GB DDR3 ram, 1TB disk, and the phenom II 2.6.

However if you are really after the performance it seems that once linux is tuned for lynnfield it should be a good bit faster, especially in the case of the i7-860. It's hard to compete with, faster single thread because of slightly higher IPC, faster single thread because of turbo boost, and faster performance for threaded apps when using hyperthreading and 8 threads (unlike hyperthreading on the p4).

So all in all you have to ask your self do you want a great desktop system for $600-$800 that will be a pleasure to use for a wide range of uses or do you want to pay another $200 ish for the i7-860.

Granted the turbo boost will take some time to fix, but I'd expect it to be very small relative to the useful life of the system.

[BillBroadley]

Could you also post some Java benchmarks? Using the sun java compiler. This I feel provides a fairly good estimate of performance as the binary is provided by sun and is always the same and I assume is very well optimized by them for multiple architectures.

Before I saw these benchmarks, I was totally kicking myself for ordering a P II X4 955 for my lab. This makes me feel much better.

I have some stream memory benchmark code that I ported to java, not sure what the point is. I've not found java's JVM particularly efficient or highly optimized, in fact just the opposite. Java has been very slow to adopt 64 bits, vector operations, hell it doesn't even implement registers. For this reason there's an industry built around doing java better than sun. Take a look at ARM (gazelle), android (android phones don't actually run java), and various others shipping java for real time or performance sensitive applications. Nor are java apps particularly common. So any resulting numbers don't tell you much about how well applications will run on it.

[djiezes]

Granted the turbo boost will take some time to fix, but I'd expect it to be very small relative to the useful life of the system.

Thanks for all your answers, that's pretty much what I needed to know.

[Kano]

"Intel Turbo Boost Technology was disabled during our testing due to the aforementioned problems." - i did not get to the point in the article where the problems are mentioned, but disabling the feature that makes the i5 a better choice than other intel quads in comparision with higher clocked duals is somehow stupid. Basically the idea behind it is really good, as when you manually oc then you would need more power for the cpu all the time and there it oc only when needed - and still does check the power consumption which is of course needed to be cooled. As Intel ships really small default coolers the limits could be more restrictive than they would be, but you always oc manually, just that this is guaranteed. Why would you use lower speed if not needed?

[justapost]

I received an i5 750 today, together with an GBT P55-UD4 mobo. I compared it with my 955BE + GBT GA-MA785GMT-UD2H. Only mobo and cpu differ between setups. Both used 4GB OCZ Plats at 1333MHz CL7 and an nvidia 8800GT 1GB gfx. As os I choose sidux 2009-2 dist-updated. I left all power saving features on and also enabled turbo on the 750. cpufreq-acpi seems to ignore the two and one core increases. The chip ran at 2.8GHz most of the time.
http://global.phoronix-test-suite.com/?k=profile&u=justapost-5168-12682-147
I plan to run the full universe suite and more clock vs. clock comparisons in the next days.

[Apopas]

I received an i5 750 today, together with an GBT P55-UD4 mobo. I compared it with my 955BE + GBT GA-MA785GMT-UD2H. Only mobo and cpu differ between setups. Both used 4GB OCZ Plats at 1333MHz CL7 and an nvidia 8800GT 1GB gfx. As os I choose sidux 2009-2 dist-updated. I left all power saving features on and also enabled turbo on the 750. cpufreq-acpi seems to ignore the two and one core increases. The chip ran at 2.8GHz most of the time.
http://global.phoronix-test-suite.com/?k=profile&u=justapost-5168-12682-147
I plan to run the full universe suite and more clock vs. clock comparisons in the next days.
Similar results with Michael's tests which show Phenom to beat i5 in general, whether the windows' benchmarks shows the opposite...

[Kano]

Well maybe you need a much better cpu cooler to enable higher turbo boost or the bios is bad. Maybe there is a newer beta one.

[justapost]

Well maybe you need a much better cpu cooler to enable higher turbo boost or the bios is bad. Maybe there is a newer beta one.
Well cpufreq-info only shows speeds up to 2.8GHz. Maybe turbo works better if I disable EIST and do not use cpufreq-acpi at all.
I tried to log the cpufrequency during the test but only one core appeared. As for cooling, I'm currently only using the stock cooler because I do not have the proper clips for my other ones. I expect an Corsair Hydro H50 in the next few days but I thought I only need it to reach ~4GHz and not at stock. :)
Can be it's a bios issue with the linux acpi tables, under windows I already saw above 3GHz.
Will do more testing tomorrow, preparing universe will take a while. :o

[AdrenalineJunky]

Well maybe you need a much better cpu cooler to enable higher turbo boost or the bios is bad. Maybe there is a newer beta one.

since you meantioned you didn't read what the problem was

enabling turbo boost led to erratic/poor performace, test results varied greatly from one run to the next and all of them were worse then with it off. also the cpu wasn't actually scaling up like it should (thats what micheal had happen any)

[Kano]

Thats not really logical to me when this would be a Win only feature. That is not really predictable that is partly clear as it depends on the core temp.

[justapost]

Updated the bios to the lates available version..
http://global.phoronix-test-suite.com/global-graph.php?g=BAR_GRAPH&t=FLAC%20Audio%20Encoding&s=WAV%20To%20FLAC&n=1.2.1&u=Seconds&i=i750%20stock%20EIST%20Off%20CS%20Off%20Turbo%20O ff;i750%20stock%20EIST%20Off%20CS%20Off%20Turbo%20 On;i750%20stock%20EIST%20Off%20CS%20On%20Turbo%20O n;&v=11.46;10.91;9.55;&p=LIB&x=2.2.0a2
Then I played with Turbo, C-State and Eist settings. For the higher turbomodes in windows all C-State options must be enabled. Without turbo behaves like I thought it behaved under linux yesterday and the max multi was 21x (2.8GHz).
Looking at the linux results it seem higher multis where in effect yesterday but the correct frequencies are not reported under /proc/cpuinfo.
With EIST/Turbo and CS off the result looks like 2.66GHz. Once I enable Turbo but leave C-States disabled it looks moore like 2.8GHz and with C-States enabled the frequency should have been 3.2GHz.

11.46*2.66GHz = 30.48
10.91*2.8GHz = 30.54
9.55*3.2GHz = 30.56

A sidenote, results are repeatable here.

[Kano]

Could you do 7zip benchmarks? I like those as they are really fast to test - maybe with EIST on too.

[justapost]

The system is busy running universe-cli here atm, but 7zip already ran.

i5 750 Turbo/CS/Eist On: 8647,66 MIPS
955BE CnQ On: 7553.00 MIPS

[Kano]

64 bit i guess, that's intersting my 3.16 s775 cpu beats that.

http://global.phoronix-test-suite.com/?k=profile&u=kano-11623-26178-18490

[justapost]

He He you ran the pts-livecd. Can it be your test file was on a ramdisk. ;)

Update: Can't be looking at the second result. 8GB may have an impact.

Found those intel results.
http://global.phoronix-test-suite.com/?k=profile&u=powtrix-22206-11719-16230

I need 3.8GHz on the 955BE with 8GB ram to reach your results.
http://global.phoronix-test-suite.com/?k=profile&u=justapost-6742-23762-13556

[Kano]

That was a test PTS vs. Kanotix 64 Excalibur. Kanotix was faster in that one, but slower in some others. Btw. i have got results with 3.8 GHz too.

http://global.phoronix-test-suite.com/?k=profile&u=kano-12032-23210-22346

http://global.phoronix-test-suite.com/?k=profile&u=kano-6627-14842-17066

To be fair, the EIST was enabled in BIOS, but Linux was set to use performance mode.

[justapost]

Thank you for the 3.8GHz data, I like comparisons.

Here are a few more results, this time universe-cli.
http://global.phoronix-test-suite.com/?k=profile&u=justapost-5358-5083-22089

Currently I run the first serie again with C-States And C-Satetes+Turbo disabled. Will have to rerun the test with both enabled because the newer bios runs with an slightly higher BCLK (~136MHz vs. ~133MHz).
One of those nas benchmarks (IS.c) cause the system to hang, once during the 955BE test and once during the i5 750 test with C-States disabled.

[justapost]

Enough for today.

http://global.phoronix-test-suite.com/?k=profile&u=justapost-18550-16695-143

I ran with Turbo and C-States enabled twice, to show the results are consistent. Also I ran with C-States disabled, which means the cpu ran at 2.8GHz and also in addition with turbo disabled.

In the comparison I used the reciprocal for time results and calculated the score increase over the 955 results for each test and average over all results.

http://www.abload.de/thumb/resultsyf3y.jpg (http://www.abload.de/image.php?img=resultsyf3y.jpg)

Overall with that selection of benchmarks an 955BE performs like an i5 750 at ~2.75GHz. I do not think this selection is representative for daily usage.

[Kano]

Povray is a nice SINGLE thread example that shows the turbo mode. Excactly 20% faster, 3200 MHz.

[justapost]

Here's an comparison against turbo off.
http://www.abload.de/thumb/results-turboch1x.jpg (http://www.abload.de/image.php?img=results-turboch1x.jpg)

[Kano]

You can be sure, everything with 20% performance boost lets the cpu run with 4 x 133 more and is most likely single threaded (look at htop). Maybe you could check cpu speed with /proc/cpuinfo too when this is running...

[justapost]

The problem is the higher multis are in use but they are noot reported.
With Turbo and c-states enabled cpufreq-info shows
2.8GHz, 2.66GHz followed by the frequencies with lower multis.
With turbo on and c-states off it shows 2.66GHz twice.
In the first case multis above 21x are in use but max 21x can be tracked. In the second case only the 21x multi is in use but even if the chip runs at 2.8GHz cpufreq-info and cat /proc/cpuinfo only show 2.66GHz. :)
I made an config for the onboard sensor chip so I can atleast log cpu voltage and temperature + cpu usage.

[Kano]

Well that's secret OC then ;) The first 133 mhz step is always on with load and good cooling. The higher states only when not all cores are used. But basically it does not really matter if you see it, at least the speed is there and it runs fast.

[justapost]

If you modify the p-states of an phenom black edition chip, an modified multi does also not show up in /proc/couinfo. :rolleyes:
Looking at apache and postgres results it seems those had problems with eist. I prepared a suite with nearly all benchmarks (I left out all disc benches and the retail ones), currently I test the pII at 2.8GHz with that suite, next will be the i5, this time with eist and c-states disabled. Should be an good clock vs clock comparison, only the uncore speed (2.13GHz( differs from the nb speed (2GHz). I wonder how long it will take to finish one run, think I can run one or two tests per day.
The NAS IC.s benchmark which caused system hangup's here works fine with increased vdimm, so it's a good and quick memory stability test.

[Kano]

There are more hidden values from cpuinfo. Like when you don't disable EIST on s775 and you go over FSB 400 then you will not see the correct speed too. Of if you like to get fake high speeds you just need to force a lower multi than the max. of the cpu. Even when you raised the FSB lower than the default it will show up to FSB 400 just max cpu multi * fsb.

[jpiszcz]

Hello,

Will new benchmarks be performed with a newer BIOS (assuming that corrects the errors in DMESG)?

I ask because in the anandtech article, the i870 CPU wins many of the benchmarks, one or two over the top of the line extreme chips.

I was wondering is the problem with Linux or is it a bios issue?

It sounds like it could not shift the CPU speed to a higher level/bin?

Justin.

[murthyna]

Phoronix: Intel Core i5 750, Core i7 870 Linux Benchmarks

Now that we have provided a brief overview of the Intel P55 and how it functions under Linux, our larger area of concentration is looking at the Linux performance of the P55 with the new Core i5 750 and Core i7 870 processors. We have a number of benchmarks in this article along with more information on these Lynnfield processors.

http://www.phoronix.com/vr.php?view=14174

I have gone through the review and it gives a good comparison. I have got a new board DP55WB with i5 750. However I could not get it to work with all cores enabled for Ubuntu 9.10 amd 64. It works only with one core enabled in the BIOS. However, the performance indicated in the review does not look like a single core's performance. Was something special done to make it work with all the cores enabled?

BTW, my m/c works fine with Windows 7 RC with all the cores enabled, so that rules out any h/w, board problems.

[murthyna]

I have got a DP55WB board with i5 750 processor. It doesn't boot Ubuntu 9.10 amd64 with all the cores enabled. It works only with one core enabled in the BIOS. However, it works fine with Windows 7 with all the cores enabled.

Was there something special done to enable all the cores in the performance tests of the review?

[Kano]

The first thing you do with a new board is updateing the BIOS ;)

[murthyna]

Thanks, I have updated the BIOS to the latest ver. Still no luck, everything remains the same. Anything else to be checked?

[ad_267]

Hmm I was looking at buying an i5, I'm not sure if that's a good idea now. From the "Another Look At Intel's Lynnfield Linux Performance" article, it looks like BIOS upgrades may solve some problems. I'm looking at getting a Gigabyte motherboard, what are people's opinion on them?

[StringCheesian]

Hmm I was looking at buying an i5, I'm not sure if that's a good idea now. From the "Another Look At Intel's Lynnfield Linux Performance" article, it looks like BIOS upgrades may solve some problems. I'm looking at getting a Gigabyte motherboard, what are people's opinion on them?

I'm pretty sure you would not need a BIOS upgrade. The articles made it sound like Michael was given a motherboard with early beta BIOS whereas motherboards you can buy now are shipping with fixed BIOS.

Here's what I'm talking about:We will be running our own tests using a newer BIOS for the Intel P55 motherboard shortly. With these Intel-provided results, they are using a newer, shipping BIOS and not an older BIOS provided to the press.

[ad_267]

I'm pretty sure you would not need a BIOS upgrade. The articles made it sound like Michael was given a motherboard with early beta BIOS whereas motherboards you can buy now are shipping with fixed BIOS.

Ah ok, thanks for explaining that. I also just read the review of the ECS Elitegroup P55H-A motherboard and there weren't any problems with the i7 except for no LM_sensors support, so it looks like I should be fine with an i5. I'll be getting all the parts on Monday hopefully, so fingers crossed it will all work well.

[murthyna]

Finally the issue is resolved. It was caused by, surprisingly by the USB mouse attached to the system. I found disabling USB legacy support in the BIOS allows the system to boot with all cores enabled. However, this deactivates the keyboard as well till the OS is loaded. Changing the mouse did the trick and it works fine. I didn't find any other problem with that particular optical mouse and use it with my other older systems having Ubuntu and Windows with no problem.

[popper]

whats with the total lack of any x264 tests on most I series benches , we always want x264 and ffmpeg tests and results, ALWAYSI SAY....