$100 AMD X3 710 outclasses i7 and i5 in Linux benches

Status
Not open for further replies.
There will always be situations which are ideal for a certain type of architecture or cache hierarchy. It might also be a Linux-specific or compiler issue, because in Windows, the Intel CPUs win. Since turbo mode doesn't work properly in Linux yet, there seems to be some issues with these CPUs and Linux.

You can't really change the fact that Intels $196 CPU is faster than every AMD CPU. This is very similar to the original Phenom 9600 launch, where Intel's $260 Q6600 beat the $283 9600.

The good news for me is that I'll soon be able to pick up the fastest Phenom II X4 for maybe $150 :) because this will force AMD to drop prices, even if they can't afford it.
 
JimmiG said:
There will always be situations which are ideal for a certain type of architecture or cache hierarchy. It might also be a Linux-specific or compiler issue, because in Windows, the Intel CPUs win. Since turbo mode doesn't work properly in Linux yet, there seems to be some issues with these CPUs and Linux.

You can't really change the fact that Intels $196 CPU is faster than every AMD CPU. This is very similar to the original Phenom 9600 launch, where Intel's $260 Q6600 beat the $283 9600.

The good news for me is that I'll soon be able to pick up the fastest Phenom II X4 for maybe $150 :) because this will force AMD to drop prices, even if they can't afford it.
Click to expand...

I have a similar plan in mind :) I just hope they *do* drop prices by a bit.
 
MargaretB said:
I have a similar plan in mind :) I just hope they *do* drop prices by a bit.
Click to expand...

I don't see what else they could possibly do. The introduction of the Athlon II X4 will make sure the Phenom II prices don't hit rock bottom lows. However they will at least need to be in line with Intels pricing, which means the Ph-II 965 will need to be priced at or below $200.
 
The synthetic king X58 makes up less than 25% a year later but AMD must immediately respond to P55 at launch prices? Especially with AM3 chips ready to go on AM2+ boards so it's not a mandatory mobo/dual-channel DDR3 update.

AMD will continue the budget/mid/high distribution of processors, with viable upgrades on existing motherboards. The lower-power or "game readiness" results have already been established against the X58, so Lynnfield is just adding on with a premium. Those looking for bragging rights or significantly better video editing/bulk processing will still choose X58, probably.

Now once P55 prices go down (although some 775 prices are staying), it will hurt more but AMD needs to release new tech periodically, anyway.
 
Last edited:
AMD is in for a fight, that's for sure. Here's to hoping I can pick up 955 or 965 in the low $100 range.
 
It should come as no surprise that Intel would do all performance testing on Windows since the majority of PC users are Windows users. Maybe I should dual-boot with a Linux distro on my gaming PC and really slim it down so when I know I'm not gonna be gaming my Phenom II X3 720 can really shine... interesting benchmarks for sure.
 
jww20 said:
Check out this review of the $100 AMD Phenom II X3 710 (Tri-core) against the i7 and i5 CPUs under Linux! Looks like Intel 'nailed it' all right... :cool:

http://www.phoronix.com/scan.php?page=article&item=intel_lynnfield&num=1
Click to expand...

????

I can't believe that's your response to Lynnfield's release!

You post results from a Linux Pro AMD Compilation and chant "We are the champions"?

Wow... just wow.

AMD has one segment left (the segment they used to occupy before the Athlon days). They can occupy the ultra cheap segment.

They're outclassed in the mainstream PC, Laptop, Netbook, High end PC, Workstation and Server markets.

That little Phoronix review doesn't change that.

AMD needs a new architecture badly.
 
The review would have made much more sense with a proper AMD X4.

JimmiG said:
You can't really change the fact that Intels $196 CPU is faster than every AMD CPU.
Click to expand...
http://anandtech.com/cpuchipsets/showdoc.aspx?i=3634&p=1
I'm talking $196. I'm talking faster than AMD's entire lineup.
Click to expand...
:rolleyes:

JimmiG said:
This is very similar to the original Phenom 9600 launch, where Intel's $260 Q6600 beat the $283 9600.
Click to expand...
How is this at all similar to the Kentsfield versus Phenom I Agena fiasco?:confused:

Configurations tested in this article:

$199 i7 920 LGA-1366
$285 ASRock X58 SuperComputer
$?? 3GB DDR3-1333MHz

$580 i7 870 LGA-1156
$200 (Intel estimated price) Intel DP55KG (P55)
$?? 2GB of OCZ DDR3-1333MHz

$179 i5 750 LGA-1156
$200 (Intel estimated price) Intel DP55KG (P55)
$?? 2GB of OCZ DDR3-1333MHz

$110? Phenom II X3 710 AM3
$95 ASRock M3A780GXH/128M
$?? 2GB of OCZ DDR3-1333MHz

Even if you replaced the Phenom II X3 710 with the X4 965 at $230 - the Phenom build would still be $54 cheaper than the i5 750 build as tested.

JimmiG said:
Since turbo mode doesn't work properly in Linux yet, there seems to be some issues with these CPUs and Linux.
Click to expand...

http://www.hardwarecanucks.com/foru...re-i5-750-core-i7-870-processor-review-9.html
The margins by which the 710 beat the i7 870 in some tests shows that a i7 870 with Turbo Boost wouldn't gain enough clockspeed boost to edge ahead of a 965 X4 (+30% clockspeed over X3 710) in those particular benchmarks.

The OP's article obviously didn't cover overclocking so it's important to remember that all the Intel chips in the review are capable of a higher final overclock/stock clock ratio.
 
Last edited:
If anything this test shows the ineffeciency of Linux...if the OP word is to be trusted on blank value, but I see this when multithread applications are run:
http://www.phoronix.com/data/img/results/intel_lynnfield/15.png

http://www.phoronix.com/data/img/results/intel_lynnfield/17.png

http://www.phoronix.com/data/img/results/intel_lynnfield/23.png

(Find more on your own)

And don't forget this tidbit:
http://www.phoronix.com/scan.php?page=article&item=intel_lynnfield&num=15
We will draw more in regards to conclusions once we have completed further Linux tests on these processors to hopefully sort out the Turbo Boost problem...
Click to expand...
 
deff said:
AMD is in for a fight, that's for sure. Here's to hoping I can pick up 955 or 965 in the low $100 range.
Click to expand...

That would be absolutely horrible for AMD. I hope that the only time we see prices that low for Ph2's are after the X6 makes it's debut.
 
Here we go with Atech trolling again -_-

Anybody that mentions turboboost doesn't seem to realize its actually called overclocking. So if you "sort out the turboboost problems" you need to OC the AMD chip.

Linux is MORE efficient than Nifty Doorways, *real* geeks have known this for years. Yes, the market share is small and getting slowly smaller, but that doesn't take away from the efficiency at all. I mean, really, do Windows users realize that Linux with Compiz Fusion enabled is by far the superior multi-tasker, looks better, and still runs retardedly more lean than Vista with Aero enabled? With Compiz Fusion running, I use less RAM than when I have XP loaded up.

To have an X3 keep up with a Core i5 quad or Core i7 quad in the majority of benchmarks is... wow. Like I said initially, its obvious Intel would go and do their testing on Windows for multiple reasons. It's smart. But seeing these benchmarks clearly shows that intelligent people who do things other than benchmark their machines (aka people that use Linux for regular tasks) get a lot more bang for the buck with an AMD chip. You know, as if the bang vs buck wasn't already their domain in the Windows Gaming market.
 
ElMoIsEviL said:
They're outclassed in the mainstream PC, Laptop, Netbook, High end PC, Workstation and Server markets.

That little Phoronix review doesn't change that.
Click to expand...

Phoronix do test other applications, unlike the majority (> 90%) who test almost exactly the same applications.
 
Can we just close this thread before it goes up in flames?
 
Interesting and surprising results...

ElMoIsEviL said:
AMD has one segment left (the segment they used to occupy before the Athlon days). They can occupy the ultra cheap segment.

They're outclassed in the mainstream PC, Laptop, Netbook, High end PC, Workstation and Server markets.

That little Phoronix review doesn't change that.
Click to expand...

Hey now, hold on a sec. There are plenty of servers running Linux, and many use the same apps Phoronix tested. This shows that there are still performance reasons to choose Opterons for servers over Xeons, depending on what you want the server to do, of course.

AMD needs a new architecture badly.
Click to expand...

True.

Atech said:
If anything this test shows the ineffeciency of Linux...
Click to expand...

Uh, no it doesn't. Both CPUs are running Linux, after all, any "ineffeciency" to be found would affect BOTH CPUs, not just Intel's.
 
LAiN_OfTheWired said:
Here we go with Atech trolling again -_-

Anybody that mentions turboboost doesn't seem to realize its actually called overclocking. So if you "sort out the turboboost problems" you need to OC the AMD chip.
Click to expand...

It's not a hardware fault, as it works fine under eg. Windows.
The problem is linux, and Turbo-bost is an auto feature, when dual or single threaded programs are run, not a overclock of the entire CPU like you seem to be suggesting.

Turbo boost:
http://www.intel.com/technology/turboboost/

You can ask AMD to implement the same thing, that makes more sense than sceewing the results artificially...
 
Atech said:
You are sayting that the X3 has turbo boost too?
Click to expand...

No. But look at the [H] review. The Core i5/i7 don't need TurboBoost to be faster than the X3 in Windows with single threaded apps. Clearly the lack of TurboBoost doesn't matter, unless you are saying that clock for clock the X3 is faster? :rolleyes:

Something ELSE is happening, something completely unrelated to the lack of TurboBoost.
 
facebook said:
Phoronix do test other applications, unlike the majority (> 90%) who test almost exactly the same applications.
Click to expand...

I run both Phenom II and i7 setups. They don't even compare when it comes to CPU intensive applications.

The only time an AMD CPU based configuration comes close is when there is something else bottle-necking performance in a system (not allowing the Core i7 or i5s to shine).

You see this in some games that are run at resolutions which max out Graphics performance.
 
ElMoIsEviL said:
I run both Phenom II and i7 setups. They don't even compare when it comes to CPU intensive applications.
Click to expand...
Have you tried vmware? databases? browsers? .NET applications or java applications with a GC.
It is true that intel perform good when the load on cache is minor. Prefetchers works in that type of load. But there are other types of software and what about running a lot of applications simultaneously
 
facebook said:
Have you tried vmware? databases? browsers? .NET applications or java applications with a GC.
It is true that intel perform good when the load on cache is minor. Prefetchers works in that type of load. But there are other types of software and what about running a lot of applications simultaneously
Click to expand...

Give me a break... Didn't you get banned once already? You sound a LOT like this guy...

http://www.hardforum.com/showpost.php?p=1033280450&postcount=37
 
longblock454 said:
Flawed and not optimized are not the same.
Click to expand...

Disabling a key feature of the i5 and then posting a bench is flawed.

This reminds me of the thread I read on AMDzone last night.
They hail that bench, and accuse ALL others of being paid by Intel:
http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=136697&st=0&sk=t&sd=a

Here is the same link, before the AMD censorship stepped in:

http://jamayer.org/junko/AMDZonecrap.htm

Link to thread on Anadtech about it:
http://forums.anandtech.com/messageview.aspx?catid=28&threadid=2333394&enterthread=y

Funny reading...denial in action.
 
i love how everyone is getting all riled up about a pile of synthetic benchmarks
 
Atech said:
Disabling a key feature of the i5 and then posting a bench is flawed.

This reminds me of the thread I read on AMDzone last night.
They hail that bench, and accuse ALL others of being paid by Intel:
http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=136697&st=0&sk=t&sd=a

Here is the same link, before the AMD censorship stepped in:

http://jamayer.org/junko/AMDZonecrap.htm

Link to thread on Anadtech about it:
http://forums.anandtech.com/messageview.aspx?catid=28&threadid=2333394&enterthread=y

Funny reading...denial in action.
Click to expand...

Without doing any homework, funny indeed!
 
Atech said:
Disabling a key feature of the i5 and then posting a bench is flawed.
Click to expand...

Which DOES NOT MATTER HERE. The Core i5 and i7 are both faster clock for clock. TurboBoost would only increase the lead the Intel CPUs should *already have*. [H]'s review didn't have TurboBoost, I guess it must be "flawed" as well :rolleyes:

Besides, no one disabled the feature. You make it sound malicious. It isn't. TurboBoost will likely very quickly get kernel support in Linux as Intel generally has pretty open documentation. The feature simply didn't have launch support - nothing new there. Neither did SSE, SSE2, etc... All of those required the OS to support them as well.

Atech said:
Listen , if the hardware works as intended under one OS and not under another, you point the finger at...the CPU?! *chough*
Click to expand...

Ah, but now you are changing the situation to fit your views. The CPU clearly worked as intended, it just didn't work as fast as expected. That is not an OS problem. It would be like a stick of RAM being 2GB in Windows and suddenly changing to 1GB in Linux. Some hardware components are simply OS agnostic, the CPU is one of those (should be, anyway). Some things like SSE and stuff do require the OS to support them, but the i5/i7 didn't add any of those (other than TurboBoost, which again, is not the culprit here). I suspect GCC compile flags could be the culprit. Perhaps the optimizations GCC does compared to, say, VC++ helps AMD more than Intel (not necessarily intentionally). Usually those optimizations only have a minor performance improvement, but it isn't entirely out of the question.
 
kllrnohj said:
Which DOES NOT MATTER HERE. The Core i5 and i7 are both faster clock for clock. TurboBoost would only increase the lead the Intel CPUs should *already have*. [H]'s review didn't have TurboBoost, I guess it must be "flawed" as well :rolleyes:
Click to expand...

[H]ard did a clock for clock comparision:
http://www.hardocp.com/article/2009/09/07/intel_lynnfield_core_i5_i7_processors/3

Not what Phoronix did:
http://www.phoronix.com/scan.php?page=article&item=intel_lynnfield&num=3

Besides, no one disabled the feature. You make it sound malicious. It isn't.
Click to expand...
Again:
"Our Linux benchmarking of Lynnfield was done when the Core i5 750 was clocked at 2.66GHz and when the Core i7 870 was running at its stock speed of 2.93GHz. Intel Turbo Boost Technology was disabled during our testing due to the aforementioned problems. When it came to overclocking these two processors we were able to push the i5 750 easily past 3.4GHz and the i7 870 to 3.8GHz, but the time we spent overclocking was rather limited due to our core focus on the Linux performance and not having much time prior to today's Lynnfield launch"



TurboBoost will likely very quickly get kernel support in Linux as Intel generally has pretty open documentation. The feature simply didn't have launch support - nothing new there. Neither did SSE, SSE2, etc... All of those required the OS to support them as well.
Click to expand...

Like I stated in my first post:
Atech said:
If anything this test shows the ineffeciency of Linux...
Click to expand...

So a bunked OS with artifically crippled hardware used in a benchmark is good for what?



Ah, but now you are changing the situation to fit your views. The CPU clearly worked as intended, it just didn't work as fast as expected. That is not an OS problem. It would be like a stick of RAM being 2GB in Windows and suddenly changing to 1GB in Linux. Some hardware components are simply OS agnostic, the CPU is one of those (should be, anyway). Some things like SSE and stuff do require the OS to support them, but the i5/i7 didn't add any of those (other than TurboBoost, which again, is not the culprit here). I suspect GCC compile flags could be the culprit. Perhaps the optimizations GCC does compared to, say, VC++ helps AMD more than Intel (not necessarily intentionally). Usually those optimizations only have a minor performance improvement, but it isn't entirely out of the question.
Click to expand...

No I am not read above...
 
facebook said:
Have you tried vmware? databases? browsers? .NET applications or java applications with a GC.
It is true that intel perform good when the load on cache is minor. Prefetchers works in that type of load. But there are other types of software and what about running a lot of applications simultaneously
Click to expand...

In each one of those scenarios the Intel Machine easily outmatches the Phenom II. You're talking about load on cache being minor which is irrelevant because of two things.

If the load on the cache was tremendous, Intel's caching technology is 4-5 times faster than that of AMD (the bandwidth) and the latency is nearly half.

If the load on the cache was large (meaning that data needed to be transmitted over the QPi bus between cores) than you have a QPi which has a lower latency than Hypertransport (2 HOPS vs. 3HOPS) and more bandwidth.

Any scenario you throw at it things don't look good for AMD. AMD needs a new architecture.
 
Atech said:
Disabling a key feature of the i5 and then posting a bench is flawed.

This reminds me of the thread I read on AMDzone last night.
They hail that bench, and accuse ALL others of being paid by Intel:
http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=136697&st=0&sk=t&sd=a

Here is the same link, before the AMD censorship stepped in:

http://jamayer.org/junko/AMDZonecrap.htm

Link to thread on Anadtech about it:
http://forums.anandtech.com/messageview.aspx?catid=28&threadid=2333394&enterthread=y

Funny reading...denial in action.
Click to expand...

Thanks for saving at least some of that, it's an amusing reminder (I was going by the handle "smyrgl" in that thread) although it's too bad there's not a full record of it.

I had no idea that TheGhost had the audacity to not only ban me (I sort of expected this) but to purge all of my posts so that only the most piecemeal quote replies are showing up (people often ignored the most damaging rebuttals). I sort of expected childish behavior from some of the posters on AMDZone, but I figured the mods or at least the admins had more sense than that. I guess what all they really want is a giant circle jerk with no dissenting opinions.
 
Status
Not open for further replies.
Back
Top