Apple Now Claiming Fastest Servers

[Nehemoth]

Nobody talk about this
http://www.amdzone.com/#4
http://www.apple.com/xserve/


The people at apple are crazy???

Can be apple suit for this???

 

[InlineFive]

If course they claim fastest servers, they need to try and boost their meager market share. However Intel will beat them in raw speed because of their Dual-3.2GHz Xeons. However if they go bashing against the Itanium (which is pure 64-Bit) then theirs might be a little faster because I believe the maximum Itanium speed is 1.5GHz. However, the Itaniums can be put into 8-processor servers and the G5s can only go Dual so in that instance the Itaniums would win by raw quantity.

Of course it's hard to tell since there arn't really good server benchmarks for Mac OS X and Windows alike.

-Por

 

[MikeMike]

Originally posted by: PorBleemo
If course they claim fastest servers, they need to try and boost their meager market share. However Intel will beat them in raw speed because of their Dual-3.2GHz Xeons. However if they go bashing against the Itanium (which is pure 64-Bit) then theirs might be a little faster because I believe the maximum Itanium speed is 1.5GHz. However, the Itaniums can be put into 8-processor servers and the G5s can only go Dual so in that instance the Itaniums would win by raw quantity.

Of course it's hard to tell since there arn't really good server benchmarks for Mac OS X and Windows alike.

-Por

pshaw. intel doesnt even have the fastest dual servers out currently.

that belongs to AMD as i believe anandtech proved a while ago.

MIKE

 

[Eug]

Originally posted by: PorBleemo
Of course it's hard to tell since there arn't really good server benchmarks for Mac OS X and Windows alike.

? You'd run Linux vs. Mac OS X.

Anyways, for stuff like BLAST, the Mac would win hands down, but that's because it's been heavily SIMD optimized for the Mac. For other stuff it depends on the system and the application, but I'd say that Linux with a top end dual x86 would win most of the time. It's interesting to note though that these Xserves are fairly price competitive though it seems.

The website though is just marketing fluff. To take it seriously (like AMDZone seems to have done) is strange. Actually, they seem like they're just whining to be quite frank.

 

[quikah]

Anadtech didn't prove anything, they didn't test with the fastest Intel parts, nor did they use a comprehensive testsuite. they performed ONE test, not very conclusive. That is like saying Nvidia is better than ATi because the Geforce FX 5950 scores higher than the Radeon 9700 in UT2003.

 

[BFG10K]

Apple make a lot of claims but most of them aren't even remotely accurate or correct.

 

[Eug]

Another interesting comparison would be the performance per watt.

The 90 nm G5 is one half the size of Prescott. Thus, overall, it should use up much less juice per server node. This is an important consideration if you're gonna cluster them.

 

[sparks]

Its just marketing. They are just trying to increase mind-share. Anyone who is in the market for servers would be foolhardy to not do an exhaustive comparison and testing of products before purchase.

 

[NFS4]

Who cares when they only have like 3% marketshare? Besides, it's all marketing anyway. I think Chris Tom put it best:

Should Apple be fabricating benchmarks making IBM look bad when they get their CPU from IBM? Fools. I've sent this news to IBM so we shall see if there is a response. Essentially Apple is trying to claim a 50% lead over IBM's Opteron based eServer. Here's predicting that Apple does not have the balls to send a Xserver out to say Ace's Hardware or Tech Report for real testing, just as they lack the correct parts to do so with their G5 workstations. Why has no one called them out on that? Why no legit G5 workstation reviews? This is probably why Intel doesn't send out Celerons for review. I find it laughable that the Opteron versus Xeon results from Ace's Hardware in no way mirror these results from Apple. How have they crippled the Opteron box? Let's hope we find out.

 

[NFS4]

Originally posted by: sparks
Its just marketing. They are just trying to increase mind-share. Anyone who is in the market for servers would be foolhardy to not do an exhaustive comparison and testing of products before purchase.

And I would like to know what Apple's share of the server market is

 

[Sunner]

Take a look at AMD's/Intel's/Apple's/everyone else's websites and everyone will have the fastest everything out there.
Anyone making an informed decision needs to look past marketing, it's not there to inform, it's there to sell.

Id love to get an Xserve to play with at work, never used one, for all I know, it might suck terribly, or it might kick ass at alot of things.
The last place on earth where I would look for advice would be AT though, there are just too many people here who base their oppinions on everything BUT facts.

 

[ultimatebob]

Originally posted by: NFS4

Originally posted by: sparks
Its just marketing. They are just trying to increase mind-share. Anyone who is in the market for servers would be foolhardy to not do an exhaustive comparison and testing of products before purchase.

And I would like to know what Apple's share of the server market is

I'd also like to know the percentage of available server software out there that has been ported to run under OS X server. I have a hunch that the number is almost as low as their marketshare.

Nice piece of equipment, but it's useless to me if I can't run MY preferred application or database server instead of the one that Apple wants me to run.

 

[Eug]

Originally posted by: ultimatebob

I'd also like to know the percentage of available server software out there that has been ported to run under OS X server. I have a hunch that the number is almost as low as their marketshare.

Nice piece of equipment, but it's useless to me if I can't run MY preferred application or database server instead of the one that Apple wants me to run.

What do you run? OS X is Unix, so a fair bit of stuff has already been ported. Even stuff like Oracle 10 which isn't certified on OS X yet is at least coming.

Then again, a lot of stuff hasn't been ported. Or if has been ported, it's still fairly new.

It really depends on what you want to run.

This is what comes standard with Mac OS X Server.

 

[addragyn]

Originally posted by: NFS4
Who cares when they only have like 3% marketshare? Besides, it's all marketing anyway. I think Chris Tom put it best:

Should Apple be fabricating benchmarks making IBM look bad when they get their CPU from IBM? Fools. I've sent this news to IBM so we shall see if there is a response. Essentially Apple is trying to claim a 50% lead over IBM's Opteron based eServer. Here's predicting that Apple does not have the balls to send a Xserver out to say Ace's Hardware or Tech Report for real testing, just as they lack the correct parts to do so with their G5 workstations. Why has no one called them out on that? Why no legit G5 workstation reviews? This is probably why Intel doesn't send out Celerons for review. I find it laughable that the Opteron versus Xeon results from Ace's Hardware in no way mirror these results from Apple. How have they crippled the Opteron box? Let's hope we find out.

Let's not & just keep on living our lives. This affects me naught - just more scrolling I have to do here on the forum.

:beer:

 

[addragyn]

Originally posted by: Sunner

Id love to get an Xserve to play with at work, never used one, for all I know, it might suck terribly, or it might kick ass at alot of things.

The tech press liked them with G4s & 10.2. G5 machines w/ 10.3 are just going to be better.

NetworkWorldFusion
PC Magazine
Federal Computer Week
InfoWorld

They are very popular in BioIt. AltiVec can can crank on some genomics code.

Something the AT crowd might appreciate. Apple gives you the option to service your servers. They have the pricey support plans but they also sell Xserve & Xserve RAID parts kits.

 

[NFS4]

While not "servers", I did find this:

http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6410

Ultimately, the Mac tied or led the PC in 19 of the 77 tests that comprise this report. Put another way, the PC was faster in 58 of 77 benchmarks.

And this is comparing desktop machines. Though we haven't published specific performance numbers for laptops in this report, we did run a Mac Powerbook and a Dell laptop through a subset of the same tests. The 15-inch Powerbook G4, powered by a G4/1.25GHz processor, was thumped in every test by a Dell Inspiron 8500 and its P4-M processor running at 2.6GHz. Until Apple is able to stuff a G5 processor inside one of its sexy Powerbook cases, portable Macs are likely to remain noticeably slower than their PC counterparts.

Laptops aside, the combination of faster hardware and better software has brought about a significant performance jump for pro digital photographers using the Mac. And while the PC is still quicker overall, the Mac is at least back in the race.

 

[drag]

Originally posted by: Nehemoth
Nobody talk about this
http://www.amdzone.com/#4
http://www.apple.com/xserve/


The people at apple are crazy???

Can be apple suit for this???

I don't see were Apple claims to have fastest computer.

If you know what Linpack is and how the g5 works then I have no doubt that these benchmarks are true.

What pisses me off is places like amdzone calling bullsh!t and having no real knowledge and understanding what they are talking about in the first place.

This is like saying that GCC benchmarks (the original intel vs g5 benchmarks) were bullsh!t because the intel proccessor scored higher speeds using ICC.

That's like saying that my computer scores higher fps in quake3 then it's much faster then your computer because your scores in UT2003 is so much lower.

I am only going to say this once and I know what I am talking about, because I've run these benchmarks several time myself on different hardware.

Go here to learn how Linpack works and how it works

Linpack is a program designed to test the floating point performance of very different peices of hardware.

Floating point proccessing power is something that the G5 has always done VERY VERY good in. That's why in programs like photoshop Mac's have always done well.

There is a reason why the G5 was choosen for the BigMac cluster. It's because it is needed to proccess large amount scientific calculations that are amlost exclusively in double point percision floating point units.

As apposed to unsigned Long, or signed Short, or Word, etc etc data units.


Basicly if those guys at the AMD fanboy site actually knew what they were talking (or more probably they know exactly what they are talking about, but hope that most people reading it don't) about they would quickly point out that these benchmarks LITTLE OR NO BEARING ON REALISTIC SERVER PERFORMANCE AT ALL.

Then they would be right.

ANyways from this Listing of a Java version of the linpack (there are lots of different versions original was written using fortron) benchmarks

# 261 Mflop/s; Other Mac OS G5 2 Ghz; 10/25/03 bill shine
# 260 Mflop/s; Mac Mac OS Powermac G5, 1gig ram. 250gig hd; 12/28/03 Mad Max
# 242 Mflop/s; Other Mac OS G5 2GHz; 10/7/03 Keith Stewart
# 241 Mflop/s; Mac Mac OS G5 dual 2.0 GHz; 10/28/03 Scott Therrell
# 238 Mflop/s; Other Mac OS Mac OS X 10.2.8 on a dual PPC970 ("G5") 2.0GHz; 10/17/03 Dan Birchall
# 236 Mflop/s; Other Mac OS PowerMac G5 Dual 2GHz; 6/24/03
# 235 Mflop/s; Other Mac OS 2GHz DP G5; 9/8/03 Eric Mockensturm
# 234 Mflop/s; Other Mac OS dual 2 ghz g5 running osx; 10/4/03

# 124 Mflop/s; Other Windows2000 dual opteron @ 1.4 ghz running Windows XP (32 bit); 6/25/03 Jared Richardson
# 198 Mflop/s; Other Windows2000 Opteron 142, DDR266, ASUS SK8N; 9/21/03 Yoichi Mizomata

Of course java is a bit slower then normal Linpack benchmarks.

 

[addragyn]

Doesn't he recommend XP for digital photography in a column he writes for Microsoft? Yes.

I don't buy the conclusion. It was meant to test workflow speed, which is basically conversion speed & capture, but it duplicated the same tests with lots of different software. The Mac won the FW capture, the PC won most of the USB2 capture. The conversion results can easily be swayed either way depending on which software you choose to test. The Bibble folks have done very good work in picking up where Canon & Nikon fall down. Also <a target=new class=ftalternatingbarlinklarge href="http://www.frostyplace.com/dcraw/] based on the Open Source dcraw.c does well. It's not hard to stack the deck when you're the dealer.

That aside I think the results that should be give the most credence are the [l=Photoshop numbers]http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6413">dcRAW-X</a>. Why not convert in PS since you'll likely be doing some work in there anyway. Plus PS does a very high quality conversion.

PS numbers - G5(2x2GHz) - 6 - 2 - 2 - Xeon(2x3.06GHz)

 

[gsaldivar]

Originally posted by: BFG10K
Apple make a lot of claims but most of them aren't even remotely accurate or correct.

 

[NightCrawler]

Originally posted by: BFG10K
Apple make a lot of claims but most of them aren't even remotely accurate or correct.

 

[Schadenfroh]

Originally posted by: BFG10K
Apple make a lot of claims but most of them aren't even remotely accurate or correct.

 

[onelin]

Originally posted by: BFG10K
Apple make a lot of claims but most of them aren't even remotely accurate or correct.

4 in a row!

 

[Sunner]

Originally posted by: onelin0

Originally posted by: BFG10K
Apple make a lot of claims but most of them aren't even remotely accurate or correct.

4 in a row!

Does it count if I replace "Apple" with "marketing departments" and repeat it again?

 

[Gunnar]

Dont you think its more important that Apple is now close enough and fast enough to warrant a look?

We havent even gotten to the point that OSX is an extremely managable OS on which to run a server to begin with.

And since we are talking about servers, where in all of this do we need to have Microsoft Windows compatiblity? (Even though OSX is one of the most windows friendly OS's out there).

I think too many people here at AT are utterly blind and foolish. Like AMD and Intel dont release suspicious performance numbers...

 

[txxxx]

Originally posted by: Gunnar
Dont you think its more important that Apple is now close enough and fast enough to warrant a look?

We havent even gotten to the point that OSX is an extremely managable OS on which to run a server to begin with.

And since we are talking about servers, where in all of this do we need to have Microsoft Windows compatiblity? (Even though OSX is one of the most windows friendly OS's out there).

I think too many people here at AT are utterly blind and foolish. Like AMD and Intel dont release suspicious performance numbers...

Many will not consider the Mac OSX platform until major, useful applications and general day to day tools and games are ported over. OS X has a tiny useful footprint for many.

To the last bit - I dont even read "AMD and Intel" performance numbers. Not many of us do, given all the additional testing.