Mac Pro To Continue Quad-Core Tradition? Woodcrest?

[Multimedia]

Get A Quad Tresmay

thank you "Ehurtley" and "multimedia" you are very helpful. I had purchased the Ibook in my Sig. just before the intel switch because I knew it would be a while for the Mac pro's, but I am going to school for media arts, and this thing makes my soul hurt, 'it's so F'''ing slow". I use CS2 and Studio MX, FCE and I have 3k waiting for this Mac Pro, I'm just itching --Itching--.
Rant over.

If there's any way you can afford any Quad - even if it's the G5 Quad refurb for $2799, you will be much happier than if you settle for a Dual Core only. Media Arts Demand Quad performance. If I didn't have this Quad, I would be incredibly upset most the time. The multitasking capablilty leaves any dual processor in the dust.

 

[treysmay]

Between the 20" acd and the computer, I think my cash will only allow me to get the bottom configuration with about 3 gigs of ram. Believe me! if i could in any possible way, shape or form; be able to afford a Quad Woodcrest or even a Quad g5 refurb, I would. I did think of using a middle model or top with my 17" NEC LCD but what the point of editing and colour correcting HD content without being able to see it.

 

[iBookCT77]

Build Your Own Tower®

Howsabout this.

Have the basic MacPro be a suped up Mac Mini...Either with a bigger power brick, or another inch or two taller to support an internal unit.

Put a special proprietary connecter (female) on the bottom of the unit.

From here... Apple could sell a few additional units.

One could be an expansion chassis that held a couple of PCI cards that basically looked like a mac mini w/o the CD slot.

This unit would have the male version of the proprietary connecter on the top and another female on the bottom.

Another unit would be media.. Either one chassis available with a popout panel or two for removeable media and another version that had a solid front, or just the former.

Support for multiple expansion chassis allowing more PCI cards (what is the theoretical max, anyone know?) as needed.

Third partys could provide additonal expansion chassis that could support additonal ports (hub) for FireWire/USB/Enet/etc. or memory card/stick readers or both.

The possiblilities are endless, and the out come would be a computer no bigger than necessary.

if done right, it would basically look like one really tall mac mini when expanded out.

Whatcha think?

-Phil

 

[Multimedia]

Buy G5 Quad + Dell Screen Cheap More RAM Later

Between the 20" acd and the computer, I think my cash will only allow me to get the bottom configuration with about 3 gigs of ram. Believe me! if i could in any possible way, shape or form; be able to afford a Quad Woodcrest or even a Quad g5 refurb, I would. I did think of using a middle model or top with my 17" NEC LCD but what the point of editing and colour correcting HD content without being able to see it.

Buy a Dell Monitor instead of Apple. They are both from the same Samsung source. Ram is Cheap. The G5 Quad will probably go down in price some more after the Quad Intels ship. If it's that tight for you, then get the G5 Quad and buy more ram when you can scrape together an extra $150 for two 1GB sticks.

 

[Multimedia]

All Quad Mac Pro Lineup Would Be So Great

Nope. First of all, PowerMacs are workstations. It would be lunacy to ship dual-core workstations, when your competitors (Dell) are shipping quads. And besides, using two different chips (Woodcrest & Conroe) on a single lineup of machines is not smart. Apple ALWAYS tries to maximise synergies (Mini was basically recycled iBook/eMac, Apple-remote is basically iPod shuffle with different internals etc.). Splitting up your workstations between two different CPU's would be bad for synergies. They might (and hopefully will) release an additional line of Workstations that run on Conroe, but PowerMac will be all Woodcrest.

Mark my words: Every PowerMac will be a quad.

Yes the bus is a bit slower. But the difference is neglible. And Woodcrest has twice as much L2-cache than G5 has, so that help in bus-utilization quite a bit. And Woodcrest has advanced prefetch-mechanism that also reduce the load on the FSB.

Gee, I hope you're right Evangelion. That would be so cool. I was thinking that the all Quad lineup wouldn't be until January. But the sooner the better even if the bottom is only with four 1.6 GHz Woody cores.

So here are the processor choices:

Xeon 5160 3.00 GHz/1333 MHz/4 MB
Xeon 5150 2.66 GHz/1333 MHz/4 MB
Xeon 5140 2.33 GHz/1333 MHz/4 MB
Xeon 5130 2.00 GHz/1333 MHz/4 MB*
Xeon 5120 1.86 GHz/1066 MHz/4 MB
Xeon 5110 1.60 GHz/1066 MHz/4 MB

Then I would guess these three in an all Quad config:

Xeon 5160 3.00 GHz/1333 MHz/4 MB x 4 cores = 12GHz To be as fast as possible.
Xeon 5140 2.33 GHz/1333 MHz/4 MB x 4 cores = 9.32GHz To Be 23% Slower
Xeon 5110 1.60 GHz/1066 MHz/4 MB x 4 cores = 6.4GHz To keep Apple's Cost Minimal on Bottom & almost 50% Slower than the top along with a slower bus. Typically crippled bottom model. But still very fast.

I imagine Apple wants the bottom Intel Quad to be faster than the G5 Quad. I have no idea how fast these Intel GHz compare to G5 GHz. I know the numbers mean nothing to compare.

But Evangelion, I'm sorry to tell you that Ryan Katz at Think Secret reports the first Mac Pro line will be like the last G5 lineup with only one Quad model above only dual core models below.

He also says the Merom MacBook Pro will come with an all new design. Looks like good things will come to those who wait. I'm hoping the new design will include the easy to upgrade HD access like the MacBook has.

 

[iGary]

Buy a Dell Monitor instead of Apple. They are both from the same Samsung source.

Is that why they replaced my broken pair with Phillips LCD's?

 

[Multimedia]

Broken Pair of What?

Buy a Dell Monitor instead of Apple. They are both from the same Samsung source.

Is that why they replaced my broken pair with Phillips LCD's?

Well I detect a tone of sarcasm in your post. But I can't tell what you arre referring to when you say "replaced my broken pair". Broken pair of what?

My understanding is that Apple and Dell 23" and 24" monitors are both based on the same exact screen made by Samsung. This has nothing to do with 30" screens. Are you trying to say that is not the case without saying so?

I can tell you that my Dell 1920 x 1200 24" and 1600 x 1200 20" Montiors are still working flawlessly after more than a year now.

 

[~Shard~]

Well I detect a tone of sarcasm in your post. But I can't tell what you arre referring to when you say "replaced my broken pair". Broken pair of what?

IIRC, iGary has (had?) a pair of 30" ACDs. Or at least 23"...

 

[treysmay]

I just checked educational pricing and I can barely afford the dual 2.3 I'm guessing they'll lower the prices by 3 or 400 dollars, but even then. even if I did not buy a moniter or ram or food for 2 months, I still couldnt afford it. It's almost 4000 dollars canadian, thats 1000 dollars over my computer 'moniter budget. I use a new dell 2007fp at work, and I thought I heard that they changed vendors, the 2005 or what every it was with the oakley symbol stand was the one that used the identical parts to the ACD, and i dont know if this is common or fixable, but at work, there are about 45 new dell displays with that new 'Y' shaped stand and they suck. the moniters wobble at the drop of the feather it's so annoying.

 

[BenRoethig]

Howsabout this.

Have the basic MacPro be a suped up Mac Mini...Either with a bigger power brick, or another inch or two taller to support an internal unit.

Put a special proprietary connecter (female) on the bottom of the unit.

From here... Apple could sell a few additional units.

One could be an expansion chassis that held a couple of PCI cards that basically looked like a mac mini w/o the CD slot.

This unit would have the male version of the proprietary connecter on the top and another female on the bottom.

Another unit would be media.. Either one chassis available with a popout panel or two for removeable media and another version that had a solid front, or just the former.

Support for multiple expansion chassis allowing more PCI cards (what is the theoretical max, anyone know?) as needed.

Third partys could provide additonal expansion chassis that could support additonal ports (hub) for FireWire/USB/Enet/etc. or memory card/stick readers or both.

The possiblilities are endless, and the out come would be a computer no bigger than necessary.

if done right, it would basically look like one really tall mac mini when expanded out.

Whatcha think?

-Phil

I think it'd make the hardcore Mac fans, Dell, and HP happy. The Hardcore mac fans would get something that looked really cool, but Dell and HP would get the professional customers Apple would have gotten.

 

[crypto7]

...
and I have 3k waiting for this Mac Pro, I'm just itching --Itching--. Rant over.

You're not alone! My 3-month CD matures just before WWDC

And I really hope you are correct, Mm...

Xeon 5160 3.00 GHz/1333 MHz/4 MB x 2 QUAD Top
Core2E X6800 2.93 GHz/1066 MHz/4 MB Dual Mid
Core2D E6600 2.40 GHz/1066 MHz/4 MB Dual Bottom

This would allow me to get close to 3.0 GHz for significantly less... but I am not as optimistic as you

 

[Anonymous Freak]

Then I would guess these three in an all Quad config:

Xeon 5160 3.00 GHz/1333 MHz/4 MB x 4 cores = 12GHz To be as fast as possible.
Xeon 5140 2.33 GHz/1333 MHz/4 MB x 4 cores = 9.32GHz To Be 23% Slower
Xeon 5110 1.60 GHz/1066 MHz/4 MB x 4 cores = 6.4GHz To keep Apple's Cost Minimal on Bottom & almost 50% Slower than the top along with a slower bus. Typically crippled bottom model. But still very fast.

I imagine Apple wants the bottom Intel Quad to be faster than the G5 Quad. I have no idea how fast these Intel GHz compare to G5 GHz. I know the numbers mean nothing to compare.

I would imagine a closer spread, and keeping the 1333 bus across the board. As for how the new Core architecture GHz compare to G5 GHz? They're pretty darned comparable. So figure equal GHz is equal performance (The new Intel procs will be a smidge better at integer, the G5 will be a smidge better at floating point, and probably a dead-heat for vector performance.)

My guess would be dual 3.0, dual 2.6, and single 2.3. (Dual meaning dual-socket/quad-core, single meaning single socket/dual-core.) The two sub-2GHz models are meant for low-power use. So if Apple moves into the 'blade server' market, I could see those finding a home there. (Mac mini form-factor as a server, anyone? Using the new 2.5", 10,000 RPM Serial Attach SCSI drive? Remove the optical drive, and you could either cram in a second processor, or a second hard drive.) An updated full-size Xserve would use the higher-speed procs, I imagine.

 

[Evangelion]

I would imagine a closer spread, and keeping the 1333 bus across the board.

Yep. I would expect 2x 2GHz Woodcrest, 2x 2.33GHz Woodcrest and 2x 3Ghz Woodcrest. Why the jump in Mhz in hi-end? Because the hi-end model has always had considerable better specs that the other two. Original G5-PM's were 1.6GHz, 1.8Ghz and 2x 2GHz. Now we have dual 2Ghz, dual 2.3Ghz and quad 2.5Ghz. Again: a large jump.

As for how the new Core architecture GHz compare to G5 GHz?

I would say that they are a bit faster than G5, clock for clock.

They're pretty darned comparable. So figure equal GHz is equal performance (The new Intel procs will be a smidge better at integer, the G5 will be a smidge better at floating point, and probably a dead-heat for vector performance.)

FP requires lots of mem-bandwidth. And that's something Woodcrest has loads of (large L2, fast RAM etc.). And it's vector-units are considerable beefed up when compared to previous Intel-CPU's So I would say that it beats G5 in floating point (including vector), while being considerably faster in integer.

 

[Anonymous Freak]

FP requires lots of mem-bandwidth. And that's something Woodcrest has loads of (large L2, fast RAM etc.). And it's vector-units are considerable beefed up when compared to previous Intel-CPU's So I would say that it beats G5 in floating point (including vector), while being considerably faster in integer.

Hrm. I had written a whole paragraph on why the G5 has more memory bandwidth, then disproved all my own points in research. G5 has a 0.5x bus, independent busses per socket, and dual-channel PC2-4200 memory (8.5 GB/s transfer rate) vs. Woodcrest's fixed 1.33 GHz bus (which is BETTER than 0.5x on procs slower than 2.66 GHz, only marginally slower than the old dual 2.7 GHz G5's 1.35 GHz bus, and better than the quad 2.5 G5's 1.25 GHz,) independent busses per socket (assuming they use the new chipset,) and dual-channel PC2-6400 memory (12.8 GB/s transfer rate.) But, general FP performance doesn't care about memory bandwidth, it's just that the most common measure of FP performance, SuperPi, DOES need memory bandwidth.

And I had also forgotten about the new vector unit improvement that essentially doubles performance. (Older Intel CPUs did 128-bit vector calculations, but split them up in to 64-bit chunks to actually process them, taking 2 clock cycles to finish a single instruction; the new Core-based chips do a full 128-bit calculation in one clock cycle.)

So, you may be right. Until we see some apples-to-apples benchmark comparisons of G5 vs. Core, we won't know for sure, but it does indeed look like Woodcrest might be faster on a clock-for-clock basis.

 

[avkills]

The quad core Intel machine had better be faster than the current quad or Apple will have a lot to explain; but considering the benches from the Core chips, I doubt we have to worry about them not being fast.

-mark

 

[Evangelion]

So, you may be right. Until we see some apples-to-apples benchmark comparisons of G5 vs. Core, we won't know for sure, but it does indeed look like Woodcrest might be faster on a clock-for-clock basis.

Well, since current Core Duo's beat G5 (even if we disregard the dual-core advantage over single-core G5) clock for clock (*, I think that it's safe to say that Dual Core 2's beat the G5 even more .

* = G5 and Opteron/Athlon64 seem to be about equal in performance. Sure, one is faster than the other in some benchmarks and vice versa. But overall they seem to be equal. And Core Duo seems to be quite competetive with A64 (even though current models are mobile-CPU's instead of desktops)....

 

[crypto7]

well, this might be a start...

"Apple is expected to capitalise on this by putting two dual-core 2.66GHz Xeons into the high-end model of the Mac Pro. Insiders who have seen pre-production versions of the new computers say that it is without doubt the most powerful desktop on the planet."

http://www.apcstart.com/site/tgaden/2006/07/732/apple-to-use-xeon-in-quad-core-desktop
Regarding the consumer perception of a small leap to which I referred in my first msg I think this would go at least some way toward addressing that... Apple is supposed to be working with Intel on some updated Velocity Engine (at least 128-bit vector permute, execute, etc.) to replace Altivec that will eventually make its way into PCs. So I hope that comes true but maybe not this iteration. But I expect more commercial code to be optimized for Intels, so we'll see if their benchmarketing holds true... If this is accurate it's not quite 3.0 but I'll take a 2.66... At least I'll know there's an upgrade path already

 

[Anonymous Freak]

As far as synthetic benchmarks go, the new Xeons blow every other processor away... I went hunting through the SPEC benchmark database, and for integer performance, the 3 GHz Xeon is FIFTY PERCENT faster than the second-place chip, a Sun chip. The actual SPECint2000 scores are 3059 for an HP with one Xeon 5160 processor, and 2057 for a Sun Ultra 40. (Heck, a 2.0 GHz Xeon scores 2120.) Note: in this test, having multiple processors doesn't actually seem to help, as the above HP has only one dual-core chip, and a Dell with two of the Xeon 5160 dual-core chips only scores 3065, or a pitiful 0.1% faster. So this is most likely even the score of just one core of the Xeon.

Floating point is odd, IBM has inconsistent results, in that the 'names' of the systems all say '1 CPU', while the specification listings often show more than that. The actual top floating point system is one that IBM says only has 1 CPU, but the specification listing says 8.

If we only compare single-chip (by specification) systems, then a An HP single-chip (dual-core) Xeon 5160 (likely the same machine as in the integer test above,) scores 3056, an IBM that the specification listing shows only 1 chip (2 cores) has it slightly below the Xeon with a 3048, and a 1.6 GHz (single-core) Itanium 2 comes in third at 2712. (This is counting architectures. There are many other Woodcrest-Xeons listed between the top HP and the Itanium.)

SPEC listings are good for comparing architectures, simply because while they do scale with multi-processor systems, they don't do so very much. So you're really comparing raw architecture more than anything, even when you're comparing a four-way system of one architecture to a one-way of another.

 

[Silentwave]

http://www.apcstart.com/site/tgaden/2006/07/732/apple-to-use-xeon-in-quad-core-desktop
Regarding the consumer perception of a small leap to which I referred in my first msg I think this would go at least some way toward addressing that... Apple is supposed to be working with Intel on some updated Velocity Engine (at least 128-bit vector permute, execute, etc.) to replace Altivec that will eventually make its way into PCs. So I hope that comes true but maybe not this iteration. But I expect more commercial code to be optimized for Intels, so we'll see if their benchmarketing holds true... If this is accurate it's not quite 3.0 but I'll take a 2.66... At least I'll know there's an upgrade path already

I'll bet anything that the top end if 2.66 will have 3.0 as a BTO option like the 2.16 was for the 15.4" MBP at first.