Where's the Hardware bottleneck in my workflow?

[Cloud9]

Gots me a MBP 2.4ghz sr, 2gb, 160gb

I use lightroom. Love it.

I keep my non working projects on externals.
i keep my current projects on my internal.
My internal tends to have 30 gb free at any given time.

I shoot weddings=big projects.

Problem=I see beach balls in lightroom. (tend to have itunes, mail, ical, and browsers open when working too.)

Will 4gb ram kill the beach balls? Do I need a faster internal drive instead?

p.s. If it is the ram can I get away with keeping all my projects on externals?

Thanks

 

[Eidorian]

Read speed from the hard drive to the RAM.

 

[Cloud9]

Does that mean I need a faster drive or a faster system bus? (macpro)

 

[pprior]

I'm not a LR shooter (aperture here) and I know it's supposed to be faster on lower end hardware than Aperture, but here's my thoughts anyway:

1) Open up activity monitor - when you're working are the cores maxed out? if so, you're processor bound and nothing other than a new laptop is going to help.

2) I would think 2GB is not enough, but again, check your free memory and swap use. Personally I noted a big difference in speed on my mac pro when I went from 2GB to 4GB (in just general system use as well as aperture) and less so when I went to 8GB.

3) In general, laptop hard drives are SUPER SLOW. this is compounded by small amounts of memory. I know little about mac laptops, but I assume the lower end models probably have 5400 or slower drives. IF you're getting a lot of disk use, then changing hard drive would likely speed things up as well. The new MBP is rumored to have all flash drive and that would definitely speed things up!

Probably not worth much, but good luck.

 

[147798]

I agree with pprior.

However, To ask where your hardware bottleneck is and why you're getting beach balls, requires the question of: what are you comparing performance to? Did you previously use LR on a REALLY high end PC or a Mac Pro before? Those would not be comparable to a MBP (or an iMac) in terms of hardware.

You can see MPB specs (or other Mac specs) on the support page for Apple. For instance, the MBP is here: http://support.apple.com/specs/macbookpro/MacBook_Pro_Late_2007.html

Between MBP and an iMac, the specs are pretty much exactly the same. EXCEPT the MBP line uses a 5400RPM drive, which is 25% slower in terms of rotation than an iMac HDD (how that plays out in data transfer rates to memory depends on a number of other things, especially where the data that you are trying to read on the HDD is physically stored. If the files are broken up, then the HDD read head has to wait a comparably longer time for new sectors to spin under the read head than on a 7200 or a 10K RPM drive). I don't think you can put a 7200RPM drive in a MBP.

So, if comparing LR performance to an iMac, the only answer that I know of is HDD is different. FSB, Memory speed, etc. all seem to be the same. A Mac Pro is altogether different (see http://support.apple.com/specs/macpro/Mac_Pro.html) and would help improve just about everything.

In short, more memory never hurts, and it's pretty cheap (newegg has a 4GB Mushkin upgrade for $139 w/a $20 mail in rebate). A Mac Pro, of course, gives you a killer platform that's far more expandable, if you can afford one (can you get me one, while you're shopping

Last note: while I have toyed w/LR, I am not a user. However, I have heard that large libraries in LR can cause a system slow down. Have you checked over on the Adobe LR forums with the same question? Maybe there are some app tweaks they can comment on?

 

[OreoCookie]

I don't use Lightroom, but Aperture. 2 GB are not enough, especially if you (1) use RAWs and (2) run Safari alongside. It helps to quit Safari before launching Lightroom/Aperture.

In any case, you can check for yourself if you need more RAM: launch Activity Monitor and have a look at Page Outs. Used RAM, Inactive RAM, etc. are not relevant. In a perfect world, you should have 0 page outs. You can live with a few. If you have a lot, you should add more RAM. Given your situation, you probably need more RAM.

As long as you have internal hd space, I don't think you need a new internal harddrive. Don't think too much about rpms, data density and rpms determine the linear read/write speed of harddrives (I've had my 7.2k 100 GB drive replaced by a 250 GB 5.4k drive). However, it's true that desktop drives are much faster than notebook drives.

The suggestion to get a desktop is worth mulling over. However, it crucially depends on how you work: I am addicted to portable computers, I want all my important files with me everywhere I go. I wouldn't want to have to keep the data in sync on two computers.
If all you need is a faster harddrive, get a SATA card for your expansion slot and an external SATA drive.

 

[147798]

The suggestion to get a desktop is worth mulling over. However, it crucially depends on how you work: I am addicted to portable computers, I want all my important files with me everywhere I go. I wouldn't want to have to keep the data in sync on two computers.
If all you need is a faster harddrive, get a SATA card for your expansion slot and an external SATA drive.

OreoCookie: I have only a desktop, but have been mulling a laptop. All things being equal, a 5400 RPM drive seems slow to me. I've talked to others who said it's not always adequate for HD video burning via firewire (data throughput isn't enough to keep up?). I was going to do some more research, but I was unaware you could hang an eSata drive off an expansion slot. That'd be great. What is the slot on the MBP that handles that expansion? What bus does that slot come off of? Thanks for any help.

 

[OreoCookie]

OreoCookie: I have only a desktop, but have been mulling a laptop. All things being equal, a 5400 RPM drive seems slow to me.

You cannot reduce the speed of a drive down to one number. The linear drive speed depends on three factors:
(1) the rpm,
(2) the data density, and
(3) the position on the platter -- the farther outside you are, the more data passes the head during one revolution.

So a 5.4k drive might have a larger linear read/write speed than a 7.2k drive. Ditto for desktop drives, there are some drives that have a similar linear read/write speed as Raptors for the same reason: large capacity drives have a much, much larger data density.

Now it should also be clear why 2.5" drives are slower than 3.5" drives: even though the data density might be comparable, the diameter of 3.5" drives' platters is much larger than that of 2.5" drives and hence, due to (3),
this leads to a faster linear read/write throughput.

The blanket statement that `5.4k drives are slower' is false.

What is the slot on the MBP that handles that expansion? What bus does that slot come off of? Thanks for any help.

It's the PCIe Express slot just below the headphone jack. E. g. Sonnet's Tempo SATA E34P Express Card works fine, but there are others, google is your friend

 

[Feverish Flux]

The blanket statement that `5.4k drives are slower' is false.

Notebook drive to notebook drive - not really false at all in most cases.

Anywho - to the OP, you need more RAM as a starter. I used to run an iMac and a MBP at 2GB with Lightroom/Aperture and it just doesn't cut it, especially if you have other programs open. 4GB kits are uber cheap right now.

A faster drive would help as well, but not sure if you want to go through the surgery of doing that for the minimal gain you might see.

 

[OreoCookie]

Notebook drive to notebook drive - not really false at all in most cases.

Re-read my post, please.
My new 250 GB 5.4k drive has a larger throughput than my old 7.2k 100 GB drive. So the OP is in a position similar to mine: upgrade an older drive with a newer one. You're comparing new drives of the same generation -- which is a different question.

 

[Cave Man]

Re-read my post, please.
My new 250 GB 5.4k drive has a larger throughput than my old 7.2k 100 GB drive. So the OP is in a position similar to mine: upgrade an older drive with a newer one. You're comparing new drives of the same generation -- which is a different question.

Not to mention the OP's drive is approaching capacity. That'll slow it down, too.

 

[compuwar]

You cannot reduce the speed of a drive down to one number. The linear drive speed depends on three factors:
(1) the rpm,
(2) the data density, and
(3) the position on the platter -- the farther outside you are, the more data passes the head during one revolution.

These are not the only factors, in fact the list is easily twice that-
(4) Number of heads/platters
(5) Interface speed
(6) Track-to-track speed
(7) Cache size
(8) Seek time
...

Re-read my post, please.
My new 250 GB 5.4k drive has a larger throughput than my old 7.2k 100 GB drive. So the OP is in a position similar to mine: upgrade an older drive with a newer one. You're comparing new drives of the same generation -- which is a different question.

You're comparing a one platter, two head drive to a two platter, four head drive without doing specific data ordering on the physical drive- that makes for a very lopsided comparison. All things considered, twice the number of heads *should* result in a faster benchmark with any modern drive/filesystem combination *especially* where the drive is empty for benchmark purposes. Running the same benchmark on a drive that's not full of contiguous free space might result in a different result.

 

[OreoCookie]

These are not the only factors, in fact the list is easily twice that-
(4) Number of heads/platters
(5) Interface speed
(6) Track-to-track speed
(7) Cache size
(8) Seek time
...

With the exception of (4) and a sufficiently fast interface (and SATA is fast enough), all the other points are irrelevant for raw read/write throughput (which is what I was specifically referring to before).

Linear read/write speed is not sufficient as a single performance characterization, though.

You're comparing a one platter, two head drive to a two platter, four head drive without doing specific data ordering on the physical drive- that makes for a very lopsided comparison. All things considered, twice the number of heads *should* result in a faster benchmark with any modern drive/filesystem combination *especially* where the drive is empty for benchmark purposes. Running the same benchmark on a drive that's not full of contiguous free space might result in a different result.

What do you mean lopsided? I just compare my old drive (100 GB 7.2k Seagate drive) to my new Western Digital drive (250 GB, 5.4k). I compare them, because these are the drives I have, not because it's a `fair' comparison. I got a 5.4k 250 GB drive, because I need the largest sensible drive that was available now, if I had the choice between a 250 GB 5.4k and 7.2k drive, I might have gone for the latter.

There also won't be a difference in result: I needed a new drive, because the old one was way too small. It was 95-98 % full most of the time and I had to store my pictures on one of my external drives. Now they are on the internal drive again and I have 125 GB to spare. I know which drive will be faster

 

[Cloud9]

Not to hijack the thread but....

So do I put the money into a hardrive (which?) or ram?

 

[Eidorian]

Not to hijack the thread but....

So do I put the money into a hardrive (which?) or ram?

Both are dirt cheap.

Even I have 4 GB of RAM.

 

[Feverish Flux]

Not to hijack the thread but....

So do I put the money into a hardrive (which?) or ram?

RAM is the easiest place to start.

 

[OreoCookie]

Not to hijack the thread but....

So do I put the money into a hardrive (which?) or ram?

Launch Activity Monitor (located in /Applications/Utilities) and tell us how many Page Outs you have. Then we can tell.

 

[ChrisA]

Not to hijack the thread but....

So do I put the money into a hardrive (which?) or ram?

As other have said. Use "Activity Monitor". Look at the number of "swap outs" ideally there should be zero but you can live with a number that is 10% oof the swap ins. Also look at memory usage but really I expect it to be "full" as mac OS is good at finding uses for RAM.

It may seem backwards at first but the ideal thing you want to see in Activity monitor is both cores "maxed out" too 100%. When you see this it means the computer is running to it's full potential and nothing short of buying a big MP will help. If it is not at 100% look at the memory usage, swap outs and the disk IO speed. A fast disk can do 32+ MB/Second. Wha is is yours doing?

That said, RAM is so cheap now there is no point in not maxing it out

The fastes disk will be a RAID 0 "striped" external drive with a FW800 interface. The best bang per buck size right now is to find a unit with two 500 GB drive inside. These are typically sold to video editors

 

[compuwar]

With the exception of (4) and a sufficiently fast interface (and SATA is fast enough), all the other points are irrelevant for raw read/write throughput (which is what I was specifically referring to before).

So it's your contention that (a) data never goes over a track boundary, (b) the drive never has to reread because of track/sector/head alignment, (c) the drive never gets data out of cache on a read before it's addressed and (d) the drive never has to seek to a specific track to start reading? Perhaps you're simply saying that you're only interested in throughput numbers where the data is less than a track in size, starts near enough to the start of the track that it fits inside the current track?

Modern drive firmware is so much a part of the process that they effect even "raw" numbers- sometimes even fudging things like synchronous write operations. To overcome this and measure true performance takes a much more complicated benchmark than simply measuring different data transfer sizes- *especially* where the biggest transfer size measured is 1M! That's not even a valid test for reading/writing JPEGs from a modern dSLR.

For what it's worth, blanket statements like "SATA is fast enough" ignore the reality that drives don't yet reach the interface speed and therefore start to advertise thier actual transfer rates as "Media transfer rates" (such as the two you pointed to, which differ according to the manufacturer's specs.) The delta between "media xfer rate" and "interface speed" is why cache size can be important.

What do you mean lopsided? I just compare my old drive (100 GB 7.2k Seagate drive) to my new Western Digital drive (250 GB, 5.4k). I compare them, because these are the drives I have, not because it's a `fair' comparison. )

I mean that the benchmark is not so useful once the drive starts to fill, of if you're moving around mostly large files as it's so obviously slanted to an empty drive and small files that the results most likely won't be meaningful once you get enough data on the drive. In fact, it appears to be about the simplest, and therefore least useful disk benchmark one could derive if the Web pages are accurate. 90% of the transfers are less than 512 K bytes.

Heck, simply doing the same benchmark with file sizes that are larger than the disk's onboard cache would give you a better idea of the performance difference when dealing with large image files.

A real disk benchmark that you could actually use to compare performance under normal load would use a different set of tests and a much different methodology. This benchmark is indicative of the performance of empty drives moving small files around- and not much else.

 

[compuwar]

Gots me a MBP 2.4ghz sr, 2gb, 160gb

I use lightroom. Love it.

I keep my non working projects on externals.
i keep my current projects on my internal.
My internal tends to have 30 gb free at any given time.

I shoot weddings=big projects.

Problem=I see beach balls in lightroom. (tend to have itunes, mail, ical, and browsers open when working too.)

Will 4gb ram kill the beach balls? Do I need a faster internal drive instead?

p.s. If it is the ram can I get away with keeping all my projects on externals?

Thanks

You're close to the 15% of free space that HFS+ wants to do its magic, if 30G is a "sometimes, sometimes not so much" thing, then more disk will help.

I think the newest MBPs will do 2x2G sticks and be fine, the older ones won't do dual-channel RAM and putting in 3G will give a slight performance hit compared to 2G in pure memory performance terms.

I'd probably try to stay away from iTunes and the Web browsers while working, they'll bog the system down in terms of memory, disk and CPU and see if that helps.

 

[OreoCookie]

So it's your contention that (a) data never goes over a track boundary, (b) the drive never has to reread because of track/sector/head alignment, (c) the drive never gets data out of cache on a read before it's addressed and (d) the drive never has to seek to a specific track to start reading?

I was specifically talking about one figure, a synthetic benchmark. Don't twist my words.

Perhaps you're simply saying that you're only interested in throughput numbers where the data is less than a track in size, starts near enough to the start of the track that it fits inside the current track?

I'm talking about a simple enough figure that excludes all this. It's a benchmark, yes, and certainly you cannot put the performance into one figure.

Again, I was answering one specific question of another member, explaining that rpms are not at all a useful measure of performance. Afterwards someone posted benchmarks that were mostly concerned with linear read/write speed. So I'm not even making the argument you insist I make.

 

[147798]

Again, I was answering one specific question of another member, explaining that rpms are not at all a useful measure of performance.

Not quite true. I posted the comment about RPMs (comment, not a question), and I said ALL THINGS BEING EQUAL faster RPM will produce faster performance in getting the files you need, especially if your files are fragmented. I'll stick by that statement ALL THINGS BEING EQUAL (which you kindly ignored).

Also, in researching MBP vs. iMac, it seems the new iMacs have a SATA300 controller vs. 150 on the MBP (according to ZDNet)

So, the OP asked if they should go to a desktop vs. laptop. You WILL see better performance with comparable OEM specs in a desktop vs. laptop because the drives are faster and the controller is faster (or data path is wider, actually).

This is, of course, tangential to the OP posting and probably doesn't help the OP if they are not technical, but then, that's part of the charm of forum boards

 

[Cloud9]

So the ram upgrade seems a no brainer. But the hard disk part is a little confusing. So I need a new internal or external? And if you are saying external, that means I can keep my projects all on the external? What should I be looking for in a hard disk?