The comparisons were done on a 3.06 GHz Xeon with a 533 MHz FSB and a 3.06 GHz Pentium 4b with a 533 MHz FSB, Intel doesn't have a 800 MHz FSB 3.06 GHz Pentium 4 (Note that 133 MHz is a pure multiplier to 3066 thus the reason it was clocked at 3.06 GHz in the first place). Look on the previous page for my estimated GCC compiled SPECmarks for the Pentium 4C systems. It's pretty much in dead heat.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
PowerMac G5: World's Fastest Desktop Computer
- Thread starter MacRumors
- Start date
- Sort by reaction score
- Status
The comparisons were done on a 3.06 GHz Xeon with a 533 MHz FSB and a 3.06 GHz Pentium 4b with a 533 MHz FSB, Intel doesn't have a 800 MHz FSB 3.06 GHz Pentium 4 (Note that 133 MHz is a pure multiplier to 3066 thus the reason it was clocked at 3.06 GHz in the first place). Look on the previous page for my estimated GCC compiled SPECmarks for the Pentium 4C systems. It's pretty much in dead heat.
hi ddtlm;
You hit on some very important issues here, but one has to realize that the bake offs are nothing more than marketing ploys. It was to Apples benefit to go up against the best from Intel and demonstrate superior performance. Opterons will be a major competitor to the G5 there is no doubt about it. The reality is that Apple has an interesting machine, the only issue is if they can deliever on time before they are passed by by the Intel world. Well that and keep to a reasonable revision schedule.
Personally this new machine both excites me and disappoints me. The technology is exciting as it for tells of even better machines. The performance issue is no longer glaring, but on the other hand is not outstanding. The disappointment comes from a couple of directions, one being that the top of the line model should have the top of the line video card configured, the other being the lack of internal drive bays.
The drive bay issue and a couple of other things has me wondering if the Work Station Mac that everyone talks about is still a possibility.
The software demonstrations were outstanding , I think when people sit back after a few day they will soon realize that some of the software inovations in Panther are just as important as the G5.
The real mysteries are the laptops and the servers. I have a good feeling that they will be introduced soon to arrvie with panther also.
Dave
You hit on some very important issues here, but one has to realize that the bake offs are nothing more than marketing ploys. It was to Apples benefit to go up against the best from Intel and demonstrate superior performance. Opterons will be a major competitor to the G5 there is no doubt about it. The reality is that Apple has an interesting machine, the only issue is if they can deliever on time before they are passed by by the Intel world. Well that and keep to a reasonable revision schedule.
Personally this new machine both excites me and disappoints me. The technology is exciting as it for tells of even better machines. The performance issue is no longer glaring, but on the other hand is not outstanding. The disappointment comes from a couple of directions, one being that the top of the line model should have the top of the line video card configured, the other being the lack of internal drive bays.
The drive bay issue and a couple of other things has me wondering if the Work Station Mac that everyone talks about is still a possibility.
The software demonstrations were outstanding , I think when people sit back after a few day they will soon realize that some of the software inovations in Panther are just as important as the G5.
The real mysteries are the laptops and the servers. I have a good feeling that they will be introduced soon to arrvie with panther also.
Dave
I conservatively estimate that Prescott will be at least 25% faster clock to clock than the Pentium 4. Back when the Northwood Pentium 4 was released, increasing the L2 cache from 256k to 512k alone improved performance 7-17% on benchmarks and 10% in SPEC and Prescott is ALOT more than just a cache increase.
The reason apple did not test against the opteron is that they wanted real competition... Before you "flame on" that same dual xeon that apple tested KICKS THE **** out of a dual 1.8 opteron.
look here
http://www6.tomshardware.com/cpu/20030422/index.html
funny thing with everyone talking about how the athalon64 will be king... It its the smaller brother to the opteron that just got its ass kicked.
Personally im not sure how a dumded down version of a chip will some how make it faster.
in some of the tests the dual opteron looses to a single 3.06
look here
http://www6.tomshardware.com/cpu/20030422/index.html
funny thing with everyone talking about how the athalon64 will be king... It its the smaller brother to the opteron that just got its ass kicked.
Personally im not sure how a dumded down version of a chip will some how make it faster.
in some of the tests the dual opteron looses to a single 3.06
Anyone remember the "MacIntel?"
http://www.pcmag.com/article2/0,4149,1007215,00.asp
Dvorak: clueless as usual.
http://www.pcmag.com/article2/0,4149,1007215,00.asp
Dvorak: clueless as usual.
No, the Opteron was ignored for the reasons that I outlined. Why do you think that Dell systems were used in the benchmarking instead of some generic reference system. Sure you could create synthetic benchmarks for an Opteron without having them widely distributed. It just wouldn't make sense in the context of a keynote address at an Apple convention. Yes they are comparing p4 systems that are out right now with what Apple will have in 2 months. But, I wouldn't expect Apple to publish synthetic benchmarks that have their projected systems against competitors projected systems. I don't think any company does that.
Is it true what I'm hearing about the delayed shipping dates for the G5? Thats like....NOT COOL! I need this machine before I go to college....comeon Apple!
Actually...I don't believe their ship-by dates....my iPod came like a week before the ship-by date.
Something else of notice....my friends Apple Loan application she filed for yesterday is going to go through today. Apparently alot of people want that G5.
Actually...I don't believe their ship-by dates....my iPod came like a week before the ship-by date.
Something else of notice....my friends Apple Loan application she filed for yesterday is going to go through today. Apparently alot of people want that G5.
Re: Re: PB G5 15" thoughts...
Ummm, the length of the instruction has nothing to do with the "bitness" of the chip. Generally the term "64-bit CPU" refers to the width of memory pointers the CPU can support.
If instruction length were the key then Intel would have had us beat to 64+ bits years ago (some of those x86 instructions can get pretty darned long!) ...
Ummm, the length of the instruction has nothing to do with the "bitness" of the chip. Generally the term "64-bit CPU" refers to the width of memory pointers the CPU can support.
If instruction length were the key then Intel would have had us beat to 64+ bits years ago (some of those x86 instructions can get pretty darned long!) ...
wizard:
Yeah I'm not saying I wouldn't do what Apple did.
noverflow:
It seems that the Operton handles GCC 3.x compiled code much better than the Xeon (as do the Pentium 3 and Athlon), and so would have almost certainly been the top performer in Apple's SPEC test. The P4/Xeon needs ICC to score higher than the Opteron in SPEC.
plumbercrackboy:
Edit: Added "64-bit" to above statement.
Yeah I'm not saying I wouldn't do what Apple did.
noverflow:
It seems that the Operton handles GCC 3.x compiled code much better than the Xeon (as do the Pentium 3 and Athlon), and so would have almost certainly been the top performer in Apple's SPEC test. The P4/Xeon needs ICC to score higher than the Opteron in SPEC.
Does this surprise you? There are lots of things that don't benefit at all from multiple CPUs, and lots of things that will be harmed by the distributed RAM on an Opteron.
plumbercrackboy:
Create benchmarks? I was speaking about SPEC all along, running SPEC on Opertons you can buy today. Those will be the G5's main 64-bit workstation compeditor, Dell or no Dell.
Edit: Added "64-bit" to above statement.
Happy camper but what about that new keyboard+mouse?
I love the new G5, the design will surely grow on me( I initially didn't like the original crt iMac either when it came out among al those serious looking grey machines). It's a functional design, with a bit of style thrown at it, but as it mostly sits under the desktop instead of on it in most cases, it's not that important.
What does reside on desktops however are screens keyboards and mice. There was talk about a new keyboard and possibly a new cordless mouse, any sighting of these?
I just had to spend my money on an expensive car repair, so I have to start saving all over again, but that means I can possibly afford myself the second or 3rd generation G5 (2.5 or 3 Ghz) See there is an upside to each downside!
Call me an optimist, Apple rules, 10% market share, could become 15% in afew years, I hope it's more, but this would allready mean adding 50%.
let's see!
I love the new G5, the design will surely grow on me( I initially didn't like the original crt iMac either when it came out among al those serious looking grey machines). It's a functional design, with a bit of style thrown at it, but as it mostly sits under the desktop instead of on it in most cases, it's not that important.
What does reside on desktops however are screens keyboards and mice. There was talk about a new keyboard and possibly a new cordless mouse, any sighting of these?
I just had to spend my money on an expensive car repair, so I have to start saving all over again, but that means I can possibly afford myself the second or 3rd generation G5 (2.5 or 3 Ghz) See there is an upside to each downside!
Call me an optimist, Apple rules, 10% market share, could become 15% in afew years, I hope it's more, but this would allready mean adding 50%.
let's see!
This discussion on Intel and AMD benchmarks is fascinating. I remain riveted, but let me just offer my 0.02 on the G5 and what I'm comparing it to. I'm comparing it to the G4 I kept on putting off purchasing as well as my G3 B/W 400, which I've been using for years and tolerating how long it took me to rip my 15 gigs worth of MP3's, touch up photos using Adobe PE, and go through my pretty large iPhoto library.
Why did I keep my G3 and not go with a G4 iMac and/or PowerMac G4? Well, because the G3 was so *&*&! expandable! Need more firewire ports? Add a card. Need more storage space for music and photos and don't want to install another HD in your rev A G3? Add a 60 GB firewire drive. Want to get the most out of OSX? Max your ram out to 1Gig. I loved my 400 Mhz G3, and I'm sure I'm going to love my 1.8 Gig G5 more than 4.5 times as much!!
I do understand that one of the only ways Apple will gain marketshare is to out-perform its windows counterparts, but let me tell you that there is some pent-up demand from a long-time Mac user who is simply happy that his existing platform will now scream for quite some time.
Why did I keep my G3 and not go with a G4 iMac and/or PowerMac G4? Well, because the G3 was so *&*&! expandable! Need more firewire ports? Add a card. Need more storage space for music and photos and don't want to install another HD in your rev A G3? Add a 60 GB firewire drive. Want to get the most out of OSX? Max your ram out to 1Gig. I loved my 400 Mhz G3, and I'm sure I'm going to love my 1.8 Gig G5 more than 4.5 times as much!!
I do understand that one of the only ways Apple will gain marketshare is to out-perform its windows counterparts, but let me tell you that there is some pent-up demand from a long-time Mac user who is simply happy that his existing platform will now scream for quite some time.
Sorry produce benchmark numbers. Bechmark that system. Test the performance of the system.
I was not talking about writing a bechmark program.
SPEC is a synthetic benchmark. It is completely theoretical. It does not completely reflect realwork performance. That is what I meant when I wrote synthetic.
Ok, I think I cleared that up.
Opterons/Xeons are not the direct competetor to the G5. Those are server processors not workstation processor. The p4/p5/Athlon XP/Athlon 64 is the direct competetor to the G5. Those are workstation/desktop/consumer processors.
This was my point all along. The Xeon was included and Opterons were not. I don't need to go over why again. The point is that it is not a direct competetor. Apple does not have a real direct competetor for anything server. If Apple wanted a real server product they would not have released the xserve with the G4 and lousy specs that they did. Hopefully they will recant of their ways and release a real server. Can anyone say Power4? That would be the cometetor to the Opteron/Xeon.
I wonder if Apple really turned off Hyperthreading on the Dell computer when they performed thoose dual tests...?
SpyMac says the G5 wont be ready before sept 1...
http://www.amdzone.com/articleview.cfm?articleid=1296&page=2
http://www.haxial.com/spls-soapbox/apple-powermac-G5/
http://arstechnica.com/archive/news/1056478171.html
SpyMac says the G5 wont be ready before sept 1...
http://www.amdzone.com/articleview.cfm?articleid=1296&page=2
http://www.haxial.com/spls-soapbox/apple-powermac-G5/
http://arstechnica.com/archive/news/1056478171.html
It all depends on which point you start from. I was saying that 18 months from now, there would need to be a 4GHz G5 or maybe a slower G6 that is 100% faster. If you look at the 18 months from before 2 days ago, the fastest G4 was 867 MHz, or only %64 faster in all that time. I agree with you though that the particular speed jump from 1.42 G4 to 3 GHz G5 will be huge and much more than 100%
Game Boxes
I'm just picturing the games running on these machines. With all of the processes even just running from the main cpu's they would sail. With the 128mb cards like the 9800's (or is it 9700 for macs I would bet you would never have to turn back to the pc. I don't play a lot of games but even on a g4 most run in a lackluster way. Now they (the G5) could be the platform to play them on.
I'm just picturing the games running on these machines. With all of the processes even just running from the main cpu's they would sail. With the 128mb cards like the 9800's (or is it 9700 for macs I would bet you would never have to turn back to the pc. I don't play a lot of games but even on a g4 most run in a lackluster way. Now they (the G5) could be the platform to play them on.
plumbercrackboy:
Not true. I'm familiar with several sub tests on the integer side of SPEC, which includes gzip, bzip2 and gcc compile times. You can find out what each sub test is at the SPEC homepage. As far as I know, there is nothing truely synthetic in SPEC.
So what you are saying is that Apple gets to compete against $1000 single-CPU machines with their expensive dual-G5, and that the expensive dual-Xeons and dual-Opterons are not invited? So what if Xeons are often found in servers, the same will be true for G5's by the end of the year. So what if it is taking forever for Opteron workstation boards to enter the market, they are on the way. Some are here already. People like me are going to be choosing between Opterons and G5's this fall, despite the artificial classes you are trying to push them into. People who need (or want) two processors worth of performance at a reasonable price can choose between Xeons, Opterons, and soon G5's.
Upgrading RAM is so expensive from Apple. It IS possible for me to get my own ram from elsewhere and install it, right?
I ask because with the education discount I get, the 2 gig G5 was only $2800 with the upgrade to the 250 gig HD, but no upgrade in RAM. If I could install it myself, I might get a G5 as early September.....
I ask because with the education discount I get, the 2 gig G5 was only $2800 with the upgrade to the 250 gig HD, but no upgrade in RAM. If I could install it myself, I might get a G5 as early September.....
Long time reader and Mac user, first time posting here.
A colleague sent me this link this morning, comments?
http://www.haxial.com/spls-soapbox/apple-powermac-G5/
MM
Mike Teezie:
Go someplace such as www.crucial.com that gaurentees that the RAM works in the system their website said it would.
Go someplace such as www.crucial.com that gaurentees that the RAM works in the system their website said it would.
Apple set some fairly interesting standards for the benchmark -- specifically including the use of GCC all around. Since SPEC benchmarks both processor and compiler performance, this was a legitimate (if unusual) thing to do.
ICC (intel's compiler) delivers higher SPEC ratings on the same systems than GCC does. Of course, Visual Age delivers higher ratings on 970 systems (check out IBM's results for the 1.8ghz 970). If you want to use the chip builder's compiler on one system, you should do the same on both systems. What I'd really like to see is a comparison of GCC on the Mac vs. Visual Studio on the Dell, since those would be by far the most common compilers used on those platforms.
What is more debatable was Apple's disabling of SSE2 on a couple of the Dell tests, and their use of a different malloc() library. I'd want to hear more information about the library (including the availability of the new malloc library, its affect on memory size, et cetera) and the intel's scores both with and without SSE2.
-Richard
- Status