Quad 3GHz Mac Pro vs Quad 2.5GHz PowerMac

[Multimedia]

Do-It-Yourself Clovertown Upgrading Will Void Your Warranty

If one were to buy a mac pro now, is the processor upgradeable to Clovertown in the future, or is that not really worth it even if it is, because you would need a faster FSB, meaning a new logic board, to take advantage of its power?

I'm sure you know this. But just a reminder that you would be dealing with an extremely fragile and tricky upgrade process that could destroy your motherboard or fry the processor without the latest cooling system from Apple. Just my own caution against attempting this. Not worth the risk I think. There will be a better video card with the Dual Clovertown Mac Pro as well as other changes to the system fixing bugs discovered between now and then. Too many changes in the works for me to want to fool with such a complex system.

 

[63dot]

damn and i wanted asia... ahh but europe wont be too bad. damn it i am 5'7 so i might end up with the short end of the stick.

we are the same height...we can call ourselves the "toxic twins"

 

[epitaphic]

That chart speaks for NOTHING. Comparing a Mac Pro to old 2004 single core Dual G5 PowerMacs is a completely irrelevant and spurious "test"

Nah, man, you're missing the point of the chart:

Thats showing that the quad core Mac Pro is essentially the same speed as dual core Mac Pro. To translate it to normal mac scenario: If apple releases a 2.66GHz Conroe iMac/Mac/whathaveyou it will be able to crunch through FCP/Photoshop/etc faster than a Mac Pro because it can use regular DDR2 and won't suffer from horrendous memory latency.

The only way a quad Woodcrest or octo Clovertown is superior is if you're doing exactly what you do: send a bunch of things to be encoded at the same time. Video editing is a small part of the professional market. People who only encode videos day in and out are an even smaller group.

 

[gnasher729]

Thats showing that the quad core Mac Pro is essentially the same speed as dual core Mac Pro. To translate it to normal mac scenario: If apple releases a 2.66GHz Conroe iMac/Mac/whathaveyou it will be able to crunch through FCP/Photoshop/etc faster than a Mac Pro because it can use regular DDR2 and won't suffer from horrendous memory latency.

It only shows that one company can expect to get massive complaints from its customers soon about its crappy software. An H.264 encoder can easily use two dozen cores if they are there (apart from the fact that it might be limited by the speed of the DVD drive if you encode straight from DVD); there is no reason at all why this software shouldn't be twice as fast on a Quad core and four times as fast on an eight core machine.

 

[milo]

So what apps will saturate all four cores or at least get close to it, on either a quad G5 or quad xeon? Are there any?

Are there any apps that really take advantage of four cores on their own?

 

[bigbossbmb]

i know that the latest version of maya takes advantage of more than two cores on its own. i would hope that compressor does, but i'm not sure.

 

[paulvee]

aaarrrrgggghhhh....

My 3.0's shipping date just changed - for no obvious reason - from 8/20 to 9/19. One month. Clearly, something just got snagged in the supply chain.

Anyone else have this?

 

[Multimedia]

Comparing One Mac Pro Pair Of Cores To A 2004 PowerMac G5 Is Absurd And Irrelevant

So what apps will saturate all four cores or at least get close to it, on either a quad G5 or quad xeon? Are there any?

Are there any apps that really take advantage of four cores on their own?

Toast 7.1 UB can use more than two cores. In my test at the Apple stopre last Saturday I saw Toast 7.1 UB use more than 3 - between 2.3 and 3.1 cores all the time on the Mac Pro. It also uses more than two on the Quad G5 - just barely. Handbrake is not yet optimized for Mac Pro and uses a little less than two on both. That use of two is negatively impacted as soon as you start doiong something else especially both Toast and Handbrake at once.

But in future it will use all four. The problem with that "test" you so highly value, is that the testers didn't have a Quad to compare to, so they didn't even search out applications that are already "Quad Core Ready" - that would make a nice bullet on a software package wouldn't it?

Better yet: "MultiCore Ready".

If you don't think you are going to ever use more than one thing at a time, then you are right. But I think most of us here have 10-15 things open at once and do all sorts of things at once. That's the reason for "Spaces" in Loepard.

 

[paulvee]

My 3.0's shipping date just changed - for no obvious reason - from 8/20 to 9/19. One month. Clearly, something just got snagged in the supply chain.

Anyone else have this?

okay, it seems to be a RAM bottleneck. I had ordered a couple of 2 gig chips from apple cause I didn't mind paying the penalty now in order not to have to sell 1 gig'ers later on.

anyway, I'm on the phone now, getting standard RAM configuration, then I'm just going to to with OtherWorld's RAM.

I wish Apple had gotten their RAM supplies in order before they started shipping. Well, what can you do.

 

[amin]

... If apple releases a 2.66GHz Conroe iMac/Mac/whathaveyou it will be able to crunch through FCP/Photoshop/etc faster than a Mac Pro because it can use regular DDR2 and won't suffer from horrendous memory latency.

Where is your evidence to back up this statement? You have presented limited evidence about FCP, and none regarding "PS/etc." So far there is no "Mac/whathaveyou." Do you think a Conroe iMac will beat a Mac Pro due to lower memory latency alone? Do you have real experience or data regarding how horrendous a problem this is? Extra dual-core processor aside, the Mac Pro has a higher speed FSB, higher memory bus bandwidth, higher RAM capacity, and ability to set up internal RAID amongst other advantages over a Conroe iMac.

 

[chatin]

okay, it seems to be a RAM bottleneck. I had ordered a couple of 2 gig chips from apple cause I didn't mind paying the penalty now in order not to have to sell 1 gig'ers later on.

anyway, I'm on the phone now, getting standard RAM configuration, then I'm just going to to with OtherWorld's RAM.

I wish Apple had gotten their RAM supplies in order before they started shipping. Well, what can you do.

I purchased Kingston PC2 5300 FB for my Mac Pro from New Egg. They seemed to have the best price and some Mac friendly reviews.

My Pro now starts 10.4.7 in less than 5 seconds!

 

[Multimedia]

Great Caesar's Ghost! OFF To Desktop in 5 Seconds? This I Have Got To See

I purchased Kingston PC2 5300 FB for my Mac Pro from New Egg. They seemed to have the best price and some Mac friendly reviews.

My Pro now starts 10.4.7 in less than 5 seconds!

Great Caesar's Ghost! 😱 From OFF? 😱 With total 3GB RAM? You have a QT Movie you can post of that? 😱

 

[081440]

My Pro now starts 10.4.7 in less than 5 seconds!

NO WAY!! that would be awesome

 

[twoodcc]

NO WAY!! that would be awesome

yeah...please share a video if you can

 

[chatin]

From the time the Apple logo is displayed. There is a pause before that starts, I'd say only 10 seconds or so.

 

[epitaphic]

Do you think a Conroe iMac will beat a Mac Pro due to lower memory latency alone? Do you have real experience or data regarding how horrendous a problem this is? Extra dual-core processor aside, the Mac Pro has a higher speed FSB, higher memory bus bandwidth, higher RAM capacity, and ability to set up internal RAID amongst other advantages over a Conroe iMac.

Obviously, inherently the iMac design is inferior to the Mac Pro/Powermac. But I think there's a bigger reason why Apple chose to go all quad with the Mac Pro: Apple chose all quad because a duo option would have had the same performance in professional apps (again, excluding handbrake and toast which are the only two examples touted about). A single processor Woodcrest or Conroe option will have the same obtainable CPU power for 90-95% of the professional market for another 6-12 months at the very least.

Here's some data regarding the Mac Pro's FSB:

the Mac Pro (...) actually takes longer to access main memory than the Core Duo processor in the MacBook Pro. This is much worse than it sounds once you take into account the fact that the MacBook Pro features a 667MHz FSB compared to the 1333MHz FSB (per chip) used in the Mac Pro.

What can we take from this? Because of the use of FB-DIMMs, the Mac Pro's effective FSB is that of ~640MHz DDR2 system.

And how does it fare in memory latency?

It's not Apple's fault, but FB-DIMMs absolutely kill memory latency; even running in quad channel mode, the FB-DIMM equipped Mac Pro takes 45% more time to access memory than our DDR2 equipped test bed at the same memory frequency.

As for bandwidth, although the Mac Pro has a load of theoretical bandwidth, the efficiency is an abysmal 20%. In real use a DDR2 system has 72% more usable bandwidth. (source here)

I don't know bout you, but if I were a heavy user of memory intensive apps such as Photoshop, I'd be worried. Worried in the sense that a Conroe would be noticeably faster.

Memory issues aside, Woodcrests are faster than Conroes, 2.4% on average (source here)

 

[epitaphic]

If you don't think you are going to ever use more than one thing at a time, then you are right. But I think most of us here have 10-15 things open at once and do all sorts of things at once. That's the reason for "Spaces" in Loepard.

We all probably have 15+ apps running at any time, but its very rare to have more than two hammering the CPU (unless its "automated" like with handbrake/toast). That is of course, unless you find yourself editing video whilst designing a website whilst laying out a book whilst writing some music whilst watching superman at the same time. 😉

 

[Multimedia]

Mac Pro Boots 15 Seconds From OFF? 3GB Ram Total?

From the time the Apple logo is displayed. There is a pause before that starts, I'd say only 10 seconds or so.

So You are saying 10 seconds from OFF to the Grey Apple then 5 more seconds to the desktop? With 3 GB of New Egg + 2GB RAM? That's still very fast. Quad G5 is almost as fast as that though.

 

[Multimedia]

Omni Optival 2GB Mac Pro RAM Kits Only $333.99 • Link Below

I purchased Kingston PC2 5300 FB for my Mac Pro from New Egg. They seemed to have the best price and some Mac friendly reviews.

My Pro now starts 10.4.7 in less than 5 seconds!

Link To New Egg Mac Pro Kingston Memory? I can't find it. How much money?

Ramseeker.com has a full Mac Pro line up of choices now. Omni OPTIVAL is only $333.99 per 2GB kit including cool blue heatsinks IF YOU USE THIS LINK.

 

[chatin]

Here is the link to the fast memory.

http://www.newegg.com/Product/ProductList.asp?DEPA=0&type=&Description=5300+fb+dimm&Submit=ENE&Ntk=all&N=0&minPrice=&maxPrice=&Go.x=0&Go.y=0

The desktop literally explodes onto the screen! The clock timer gets only one quarter the way around one rotation. I'll see if I can shoot a quicktime movie for future Quad G5 switchers.

🙂 😛

 

[shawnce]

So You are saying 10 seconds from OFF to the Grey Apple then 5 more seconds to the desktop? With 3 GB of New Egg + 2GB RAM? That's still very fast. Quad G5 is almost as fast as that though.

A little random trivia I learned at WWDC 06...

- When you see the Apple symbol on an Intel Mac that means EFI boot loader is active.
- When you see the spinning progress indicator that means the kernel has taken over from EFI boot loader.
- When you see the the switch to blue with progress dialog then the logininwindow is active (launchd has been loading required boot time services by this point).

If you hold down option while booting and get into the traditional boot disk selection screen on a Intel based Mac you can add and remove storage devices and they will appear/disappear automatically (EFI allows for much faster scanning and dynamic add/remove of devices). Additionally they will use the volume icon if one is found and for fun you can use your IR remote to make your boot selection.

 

[amin]

Obviously, inherently the iMac design is inferior to the Mac Pro/Powermac.

It may be obvious, but based on your earlier statement that a Conroe iMac would be "able to crunch through" apps faster than a Mac Pro, the obvious seemed worth identifying.

But I think there's a bigger reason why Apple chose to go all quad with the Mac Pro: Apple chose all quad because a duo option would have had the same performance in professional apps (again, excluding handbrake and toast which are the only two examples touted about). A single processor Woodcrest or Conroe option will have the same obtainable CPU power for 90-95% of the professional market for another 6-12 months at the very least.

So you think they put an extra processor in across the line just to be able to say they had a quad? Even the AnandTech article you used as a source showed here that PS took advantage of quad cores in Rosetta

Here's some data regarding the Mac Pro's FSB:
*snip*
What can we take from this? Because of the use of FB-DIMMs, the Mac Pro's effective FSB is that of ~640MHz DDR2 system.

And how does it fare in memory latency?
*snip*

Your points about latency and FSB are not separate negatives as you have made them. They are redundant theoretical concerns with implications of unclear practical significance.

As for bandwidth, although the Mac Pro has a load of theoretical bandwidth, the efficiency is an abysmal 20%. In real use a DDR2 system has 72% more usable bandwidth. (source here)

I don't know bout you, but if I were a heavy user of memory intensive apps such as Photoshop, I'd be worried. Worried in the sense that a Conroe would be noticeably faster.

I am not worried. Everything anyone has come up with on this issue are taken from that same AnandTech article. Until I see more real-world testing, I will not be convinced. Also, I expect that more pro apps such as PS will be able to utilize quad cores in the near future, if they aren't already doing so. Finally, even if Conroe is faster, Woodcrest is fast enough for me 😉.

Memory issues aside, Woodcrests are faster than Conroes, 2.4% on average (source here)

I think you misread that. They were comparing Core 2 Extreme (not Woodcrest) and Conroe to see whether the increased FSB of the former would make much difference.

 

[Multimedia]

Wow! 2GB Sticks For Only $349 Each? That Sounds Like A New Low Price

Here is the link to the fast memory.

http://www.newegg.com/Product/ProductList.asp?DEPA=0&type=&Description=5300+fb+dimm&Submit=ENE&Ntk=all&N=0&minPrice=&maxPrice=&Go.x=0&Go.y=0

The desktop literally explodes onto the screen! The clock timer gets only one quarter the way around one rotation. I'll see if I can shoot a quicktime movie for future Quad G5 switchers.

🙂 😛

Wow! 2GB Sticks For Only $349 Each? That Sounds Like A New Low Price. Thanks for the video. I hope I can hold out for the Conroe-Merom roll outs before I pull the trigger.

 

[Half Glass]

Does this bother anyone else? MultiCore Ready FCP?

"Quad Core Ready" - that would make a nice bullet on a software package wouldn't it?

Better yet: "MultiCore Ready".

So the webpages at Apple.com suggest the improvement of Xeon vs Quad G5 in FCP of 1.3- 1.4 times as fast as the Quad G5.

However, notice that it is footnoted that these results were obtained using a Beta version of FCP:

Testing conducted by Apple in July 2006 using preproduction Mac Pro units and shipping Power Mac G5 Quad units. Mac Pro testing conducted with a beta version of Final Cut Pro.

What does this mean for us: new version of FCP required for multiple core utilization?

I just took advanatage of the crossgrade, will it soon need an upgrade as well? I hope it is an update/download not as a version upgrade/purchase.

--HalfGlass

 

[epitaphic]

So you think they put an extra processor in across the line just to be able to say they had a quad? Even the AnandTech article you used as a source showed here that PS took advantage of quad cores in Rosetta

Yes under some specific results the quad was a bit faster than the dual. Though with the combo of Rosetta+Photoshop its unclear what is causing the difference. However, if you compare the vast majority of the benchmarks, there's negligible difference.

Concerning Photoshop specifically, as can be experienced on a quad G5, the performance increase is 15-20%. A future jump to 8-core would theoretically be in the 8% increase mark. Photoshop (CS2) simply cannot scale adequately beyond 2 cores, maybe that'll change in Spring 2007. Fingers crossed it does.

Your points about latency and FSB are not separate negatives as you have made them. They are redundant theoretical concerns with implications of unclear practical significance.

I beg to differ. If an app or game is memory intensive, faster memory access does matter. Barefeats has some benchmarks on dual channel vs quad channel on the Mac Pro. I'd personally like to see that benchmark with an added Conroe system. If dual to quad channel gave 16-25% improvement, imagine what 75% increase in actual bandwidth will do. Besides, I was merely addressing your statements that Woodcrest is faster because of its higher speed FSB and higher memory bus bandwidth.

I am not worried. Everything anyone has come up with on this issue are taken from that same AnandTech article. Until I see more real-world testing, I will not be convinced. Also, I expect that more pro apps such as PS will be able to utilize quad cores in the near future, if they aren't already doing so. Finally, even if Conroe is faster, Woodcrest is fast enough for me 😉.

Anandtech, at the moment, is the only place with a quad xeon vs dual xeon benchmark. And yes, dual Woodcrest is fast enough, but is it cost effective compared to a single Woodcrest/Conroe? It seems that for the most part, Mac Pro users are paying for an extra chip but only really utilizing it when running several CPU intensive apps at the same time.

I think you misread that. They were comparing Core 2 Extreme (not Woodcrest) and Conroe to see whether the increased FSB of the former would make much difference.

You're absolutely right about that, its only measuring the improvement over increased FSB. If you take into account FB-DIMM's appalling efficiency, there should be no increase at all (if not decrease) for memory intensive apps.

One question I'd like to put out there, if Apple has had a quad core mac shipping for the past 8 months, why would it wait til intel quads to optimize the code for FCP? Surely they must have known for some time before that that they would release a quad core G5 so either optimizing FCP for quads is a real bastard or they've been sitting on it for no reason.