Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Hardware prices (ram, hd's, etc) always decrease. It never makes sense to buy now for what you need later. Apple and other companies depend on people like you to make their profit, which btw, makes it cheaper for the rest of us, so thanks!
:)
How does it not make sense to buy a computer with 8 RAM slots instead of 4 if RAM prices are always decreasing?
 
How does it not make sense to buy a computer with 8 RAM slots instead of 4 if RAM prices are always decreasing?

Wow: you're willing to pay quite a premium for some empty slots. Seems kinda wasteful to me. Seems like the money is better spent on other things (SSD boot disks, Raids/Raid cards, larger individual ram cards in order to still leave room to expand, etc). And at a certain point--and I think this is that point--you have to balance the money you're spending on a computer with the money you could be spending on other parts of your biz (cameras, talent, marketing, travel, etc.).

If the only goal is to have a screaming fast mac then I suppose you're right. Personally my goal is more project oriented.
 
"She would never say where she came from
Yesterday don't matter if it's gone
While the sun is bright
Or in the darkest night
No one knows
She comes and goes"

"Goodbye rumored Tuesday,
No updated Mac on you.
When specs change with every new day
Still I'm gonna miss you."

Apologies to the Stones:D
 
"She would never say where she came from
Yesterday don't matter if it's gone
While the sun is bright
Or in the darkest night
No one knows
She comes and goes"

"Goodbye rumored Tuesday,
No updated Mac on you.
When specs change with every new day
Still I'm gonna miss you."

Apologies to the Stones:D

Wow, great lyrics! I didn't think anyone remembered that song....

And if this waiting keeps up, the Mac Pro will soon be as old!
 
DDR2 and DDR3 prices have increased.

DD2 was chump change in late 2008 until mid 2009. Now it's up over US$100 along with DDR3. A 5x increase for the Crucial Ballistix I bought in 2008 and a smaller 33% jump for my DDR3 in September.

Sad but true. I wish I would have bought 12/16GB of DDR3 when I purchased my MP in 2009 since the prices have almost doubled. (I got 8GB for about $175 and now it's $279.)
 
Wow: you're willing to pay quite a premium for some empty slots. Seems kinda wasteful to me. Seems like the money is better spent on other things (SSD boot disks, Raids/Raid cards, larger individual ram cards in order to still leave room to expand, etc). And at a certain point--and I think this is that point--you have to balance the money you're spending on a computer with the money you could be spending on other parts of your biz (cameras, talent, marketing, travel, etc.).

If the only goal is to have a screaming fast mac then I suppose you're right. Personally my goal is more project oriented.

I see your small point but it's silly to say that 4 RAM slots is a good thing so that you can "save" your money to buy other stuff like an SSD? Are you kidding? :confused:

Do you really think that Apple lowered the price of the SP 2009 MPs with 4 slots compared to the SP 2008 MPs with 8? You're off your rocker! They saved money by giving us 4 slots then charged the same if not more.

8 slots isn't a waste, it's a way to buy the amount of RAM you can afford now and then add more later, when you can or need it without throwing any DIMMs away. Super high density DIMMs are expensive. Throwing away 1GB chips to make room for 2 or 4GB chips is much more of a waste!
 
Long time reader, first time poster :)

I found the discussion on the previous page regarding configuring the mac pro for what your needs are quite interesting. Hope you don't mind giving me some advice if I specify my needs.

I have recently gotten into Bioinformatics, which deals with processing huge amounts of DNA sequences and extracts useful information for biologists. This is (from what I can tell) quite RAM and CPU intensive. I also like playing games and want a computer that I can upgrade in the future to do so which is why I'm after the Mac Pro. Finally, I want a computer that will last me a long time, ideally 8 years.

I'm obviously not going to buy a current Mac Pro but if we bring the current Mac Pro's forward, ie imagine the new ones are 6 (SP) core for base model and 12 (DP), which one would better suit my needs? I know that having many cores can be useless if they are not all used, but the current trend in software is moving towards multi-core support and I don't see this changing. Hence, I am tending to lean towards the dual processor machine for a) more power for Perl and b) a longer-lasting computer since RAM and GPU can be upgraded but CPU cannot.
 
Long time reader, first time poster :)

I found the discussion on the previous page regarding configuring the mac pro for what your needs are quite interesting. Hope you don't mind giving me some advice if I specify my needs.

I have recently gotten into Bioinformatics, which deals with processing huge amounts of DNA sequences and extracts useful information for biologists. This is (from what I can tell) quite RAM and CPU intensive. I also like playing games and want a computer that I can upgrade in the future to do so which is why I'm after the Mac Pro. Finally, I want a computer that will last me a long time, ideally 8 years.

I'm obviously not going to buy a current Mac Pro but if we bring the current Mac Pro's forward, ie imagine the new ones are 6 (SP) core for base model and 12 (DP), which one would better suit my needs? I know that having many cores can be useless if they are not all used, but the current trend in software is moving towards multi-core support and I don't see this changing. Hence, I am tending to lean towards the dual processor machine for a) more power for Perl and b) a longer-lasting computer since RAM and GPU can be upgraded but CPU cannot.
You need to find out the specifics of the software you're using. Presumably it'll be multi-threaded, but is it fixed, or variable (runs on n cores)?

Without that, it's hard to say whether or not you'd benefit more from a hex core or a dodeca based system. Budget requirements would also help, as upgrades can sometimes be more beneficial than core counts or even clock speeds (i.e. when the cores are starved for data).

Sorry I can't help further, but those details will be the difference between getting you what you need or perhaps over-spending for a system that won't offer any benefit for the additional funds.
 
You need to find out the specifics of the software you're using. Presumably it'll be multi-threaded, but is it fixed, or variable (runs on n cores)?

Without that, it's hard to say whether or not you'd benefit more from a hex core or a dodeca based system. Budget requirements would also help, as upgrades can sometimes be more beneficial than core counts or even clock speeds (i.e. when the cores are starved for data).

Sorry I can't help further, but those details will be the difference between getting you what you need or perhaps over-spending for a system that won't offer any benefit for the additional funds.

Thanks for the reply. From what I understand, I can fork a process as much as I like therefore being able to utilise n cores. I am just getting into this field and learning more rapidly. I also know that BLAST (sequence alignment program) can make use of many cores.

Budget isn't too much of a concern, I can afford the dual-processor machine but have little money left to do upgrades after that. I just want something that can last me a long time.

This will primarily be a home computer rather than something serious I make money from (at least in the beginning). I dislike not being able to upgrade my iMac and would like space for hard drives and RAM etc. I realise some may say it's a silly thing to buy a Mac Pro for that kind of usage but I like having the power if I need it and don't mind paying for it.
 
Thanks for the reply. From what I understand, I can fork a process as much as I like therefore being able to utilise n cores. I am just getting into this field and learning more rapidly. I also know that BLAST (sequence alignment program) can make use of many cores.

Budget isn't too much of a concern, I can afford the dual-processor machine but have little money left to do upgrades after that. I just want something that can last me a long time.
What kind of disk activity occurs under this application?

(I looked at some of the applications via Wiki, and BLAST was listed, but I didn't see anything on core usage). I did notice some of the applications may be able to utilize GPGPU processing however, and if so, would be a benefit to you. So I'd research out what cards are used/best suited (i.e. are they using CUDA or other, presuming it's not written to use Open CL, which is rather new for OS X).

I'm just trying to narrow it down to either a hex or octad system, as a dodeca is going to be rather pricey. You'd need funds for additional RAM, and perhaps a disk subsystem to increase througput there (feed the cores, if it's data intesive, and given it's using genome data, I expect this is the case).
 
What kind of disk activity occurs under this application?

(I looked at some of the applications via Wiki, and BLAST was listed, but I didn't see anything on core usage). I did notice some of the applications may be able to utilize GPGPU processing however, and if so, would be a benefit to you. So I'd research out what cards are used/best suited (i.e. are they using CUDA or other, presuming it's not written to use Open CL, which is rather new for OS X).

I'm just trying to narrow it down to either a hex or octad system, as a dodeca is going to be rather pricey. You'd need funds for additional RAM, and perhaps a disk subsystem to increase througput there (feed the cores, if it's data intesive, and given it's using genome data, I expect this is the case).

http://www.perlmonks.org/?node_id=687538
This addresses multi-core perl and BLAST.

As for hard drive usage, I'm not sure yet how much it uses but I can tell you that the first human chromosome is ~300mb as unzipped text. It is the largest and there are 24 more which get smaller and smaller. So if I were analysing the entire human genome it wouldn't be quite that much data. I would say RAM is more important. On my current computer it takes a few minutes to load in the first chromosome into RAM (and for some reason it uses a gig or so of RAM, but this could be bad coding on my part).
 
http://www.perlmonks.org/?node_id=687538
This addresses multi-core Perl and BLAST.
I'd go with a hex core at a minimum, but without budget, I'm not sure an octad would be possible or not (including the cost of upgrades)

As for hard drive usage, I'm not sure yet how much it uses but I can tell you that the first human chromosome is ~300mb as unzipped text. It is the largest and there are 24 more which get smaller and smaller. So if I were analysing the entire human genome it wouldn't be quite that much data. I would say RAM is more important. On my current computer it takes a few minutes to load in the first chromosome into RAM (and for some reason it uses a gig or so of RAM, but this could be bad coding on my part).
So the sizes aren't that big, but how are they accessed?

Each as a single large file (all of it loads sequentially), or is it pulled into RAM in pieces (i.e. specific text is loaded rather than the entire file)?

This would have a bearing on determining if the access methodology would be characteristic of sustained (sequential) or random access (I'm thinking the latter), given the load times. As per bad code or not, you'd have to figure that one out.

Here's the basis for my current thinking:
300MB file size (assuming a 100MB/s file = 3 sec. sequentially,for a typical HDD; no other files to be loaded). But for random access, it could be as little as ~10MB/s = 30 sec. at best. Since there's other files likely loaded (and possibly unloaded - RAM and/or cache flush -), that time could be increased, and why you're seeing a couple of minutes for load time.

If it is in fact random access, an SSD would provide much faster performance ( SSD's aren't meant for high write environments, so it's best to keep them limited to OS/application disks for now, and write data to mechanical).
 
I'd go with a hex core at a minimum, but without budget, I'm not sure an octad would be possible or not (including the cost of upgrades)


So the sizes aren't that big, but how are they accessed?

Each as a single large file (all of it loads sequentially), or is it pulled into RAM in pieces (i.e. specific text is loaded rather than the entire file)?

This would have a bearing on determining if the access methodology would be characteristic of sustained (sequential) or random access (I'm thinking the latter), given the load times. As per bad code or not, you'd have to figure that one out.

Here's the basis for my current thinking:
300MB file size (assuming a 100MB/s file = 3 sec. sequentially,for a typical HDD; no other files to be loaded). But for random access, it could be as little as ~10MB/s = 30 sec. at best. Since there's other files likely loaded (and possibly unloaded - RAM and/or cache flush -), that time could be increased, and why you're seeing a couple of minutes for load time.

If it is in fact random access, an SSD would provide much faster performance ( SSD's aren't meant for high write environments, so it's best to keep them limited to OS/application disks for now, and write data to mechanical).

Well, an octad is possible at current prices, I'm sure it won't get more expensive? I'm happy to get an octad now without upgrades and buy upgrades as I need them later. Since it's a system for a long time I want to set it up as high as possible for non-upgradeable parts so I don't have to buy a new Mac Pro later.

The files are downloaded from the NCBI as single files per chromosome, i.e. 300mb and down to maybe 100-150mb for the smallest chromosomes. The perl script loads the entire file into an array.

I agree, I think the current long load times on my iMac are due to the OS unloading other things from RAM to make room for Perl.

Thanks for all your help :)
 
Well, an octad is possible at current prices, I'm sure it won't get more expensive? I'm happy to get an octad now without upgrades and buy upgrades as I need them later. Since it's a system for a long time I want to set it up as high as possible for non-upgradeable parts so I don't have to buy a new Mac Pro later.
Understandable. :) Though I'm not sure if it will make it for the same usage for 8 years, unless the software's not updated much between now and then. I'm more accustomed to 3 - 5 years, depending on what it's used for. Then "downline" the system to another, less demanding usage. You could potentially even place it in a cluster.

At any rate, it's a good idea to go with the highest clock speed within budget (but make sure you've current pricing at the time of purchase for RAM and other upgrades you place into the system).

The files are downloaded from the NCBI as single files per chromosome, i.e. 300mb and down to maybe 100-150mb for the smallest chromosomes. The perl script loads the entire file into an array.

I agree, I think the current long load times on my iMac are due to the OS unloading other things from RAM to make room for Perl.
You could try using a single mechanical disk, and see how it performs. Then if necessary, make the necessary upgrades to improve disk performance.

Also, I've no idea what kind of redundancy or availability you may need, given what the system is to be used for (data security and uptime respectively that can allow the system to run, even in the event of a disk failure). I'd need further information to help you with this (if you deem such a data system necessary). This would be accomplished with a RAID card more than likely, though OS X is capable of doing 0/1/10. But parity does require a card.

RAID or not, I'd recommend getting a good UPS, preferably an Online unit, such as a 1500VA unit from APC. You can get them refurbished (saves a notable bit of money), and it will protect the system from power related damage (brown-outs) as well as allow you to shut down safely in the event of a blackout.

Worst case, you could opt for a Line Interactive unit, but it's not as good, nor will it last as long (Online units should last 10yrs or so, before the capacitors go, which can be replaced). It's one of the reasons they're rebuilt and sold as refurbished units, and why there's companies that specialize in this.

Battery replacement is typically 3 - 5 years (I plan on 3 myself), and it's much cheaper to replace them than get a new unit. The Line Interactive are more of a disposable unit if it goes, but it's also cheaper to replace batteries with these as well, assuming it's still functional.

Use a good surge suppressor between the wall and the UPS, as unfortunately, the UPS's surge suppression is typically ~10% on Line Interactive units, slightly higher with the Online units (i.e ~400 J vs. 480 J respectively) of what you need (~4k Joules). Tripp Lite would be a good brand (i.e. IsoTel/IsoBar lines).

I realize this will deduct from the system budget, but it's a seriously good thing to have, assuming you don't already own such items. They're cheaper than a new system, and you won't have the down-time if the system has to go in for repairs due to power related damage (and it's possible they'd figure it out, and deny warranty coverage).
 
All aside, I'm getting a sinking feeling that there will be no new Mac Pro through July. I hope I'm wrong . . .
 
Just hold out for WWDC. If WWDC passes without a word, then it's time to give up all hope.

All of these selections of arbitrary dates upon which to give up all hope are goofy. The dates are picked and then something to prop up the rationalization is stuck to them to give them substance. Or they are picked because it will be yet another Tuesday that is largely unclaimed so far. It has been a long while since Apple purposely lined up a Mac Pro class box to a "max hype" event (e.g., WWDC, MacWorld, etc. )



For the current MacPros, Apple uses 3 different 3500 chip packages to deliver the quad versions. Right now Intel only offers one 3600 chip packages at the high end of the old 3 chip package line up. In other words, there is a gap in the component line up if Apple's plan is to move along the 3 natural progress lines and keep the prices constant. (which is exactly what they have done for years, so it is not new behavior).

Could Apple dribble out a limited number of updates? While that wouldn't hurt much financially, it is just not the regimented way they have done things over last decade. So not very likely. Just because PC vendor foo did something doesn't mean Apple is going to do it too. There will always be more configuration and broader utility offerings on the WinPC market. It is more open and Mac is closed.

No steady parts stream, no updates .... how much more simple alternative explanation could it get?

For Mac Pros ( and XServes ) more than many other boxes, Intel's schedule is Apple's schedule.
 
All of these selections of arbitrary dates upon which to give up all hope are goofy. The dates are picked and then something to prop up the rationalization is stuck to them to give them substance. Or they are picked because it will be yet another Tuesday that is largely unclaimed so far.

A tuesday within a couple of months after Intel release replacement processors for the previous line. It is logical to think this will be before WWDC as that will be focused on mobility. The problem is people saying "next week", when if we'd known WWDC dates a long time ago the saying would probably have been "by the end of WWDC".


For the current MacPros, Apple uses 3 different 3500 chip packages to deliver the quad versions. Right now Intel only offers one 3600 chip packages at the high end of the old 3 chip package line up. In other words, there is a gap in the component line up if Apple's plan is to move along the 3 natural progress lines and keep the prices constant. (which is exactly what they have done for years, so it is not new behavior).

The 3500 range is current at the two lower price points. With 2.80GHz and 3.20GHz models replacing those currently used by Apple. Other vendors don't seem to be having trouble with supplying processors launched 2 months ago. I'm sure there are reasons as there is more than just components being available involved in a product launch, but I don't think supply of processors is an issue, unless it is to do with pricing deals.
 
With 3.3GHz 6-core CPU's around the corner, you're going to have a tough time convincing me that a machine clocked at 2.26GHz is built for the future - even if it has 8 cores.

It is unlikely that Apple's refresh will consist entirely of 6 core packages.
The 5620 comes in at 2.4 GHz not 2.2. And the 3600 that is closer to the 5620 in price won't be in the 3.3GHz range. You're mixing comparisons of $1,000 processors with $400 ones. When you take into account Price/performance, it is not as clear highest GHz is always the biggest bang for the buck. Where the single package CPU costs more than twice as much it may make sense in many cases.

Not sure why Intel capped the 5520 with a somewhat too low 2.26. Perhaps to get more folks to buy off on some more expensive 3600 options at higher GHz.

Most likely Apple's base models in both the "single" and "dual" package line up will have 4 cores, not 6 .
 
In all seriousness, what percentage of updates by your account have had evidence rumors backing them up?

Haha, I was joking. :) We all know that nobody here has any super secret information. It's all just a big waiting game.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.