I'm planning on getting a new MacBook Pro soon and I was wondering what hardware upgrades I should consider getting to increase PT3 performance since PT3 is currently the slowest program I use.

Options are to upgrade the processor from 2.8GHz Intel Core 2 Duo to a 3.06GHz Intel Core 2 Duo,
double the memory from 4GB to 8GB and/or upgrade the HD from 5400rpm to 7200 or solid state. Would any of these upgrades have much effect on the performance of PT3?

Note I will be using PT3 in Windows 7 using Boot Camp, so not the mac version.

Never ran PT3 via bootcamp but other intense programs. You can't go wrong with the cpu upgrade and certainly the max RAM will be very important. The SSD would make things hum with a few caviets. If you choose to go SSD do it as a factory install rather than doing it after the fact. There might be issues with the bootcamp/W7 install due to the way W7 uses sectors/cylinders. Also, the SSD would have to be rather large (and expensive) for your DBs unless you have them set to external HDs. That said - go for the 7200rpm. Make sure they are not giving you the T6600 cpu - it has issues with virtualization.

Boot Camp isn't a virtualization tool, it lets you boot directly into Windows.

The max SSD size they currently offer is 256GB, so I'm considering holding out for a MacBook update on the chance that they will be upping this. Do you think SSD would be a significant improvement in speed? I'm guessing a lot of the slowness of PT3 comes from the DB reads, or is it mainly a memory issue?

Yup BC isn't virtualized but if you ever had to run W7 pro in XP mode (which will become increasingly less important) it would be an issue.

I don't see larger than 256 too soon - at least not at a reasonable cost. It also makes a huge difference what brand they will use - which they probably wont tell you. For instance, some laptops are coming with Imitation brand SSD which are total crap. Also, 2 - 3rd gen will be much improved and larger. I run a copy of PT3 on an SSD with 8GB RAM with an I7 820 cpu and it's crazy fast. I couldn't say exactly how much better it runs on SSD but I'm sure there is +EV performance added. My DBs are on a seperate 500GB 7200 w/32MB cache HD. I went with the SSD more for power consumption reasons (its a laptop) and the coolness factor than performance. If I were you I'd do a little research on whether there will be issues with BC/W7 on the SSD. My experience has been that if W7 is not done as a fresh install on SSD (which BC/W7 will not be) you're just asking for trouble such as stuttering and lag plus graphic issues.

Hmm thanks for the info. It sounds like there are just too many potential drawbacks to justify the rediculous price at half the size for the SSD. I think I'll just go with the 7200rpm and the upgraded CPU.

Thanks for your help

You'll be fine with the 7200. Make sure you pack that baby full of RAM - it'll be well worth it. If you need any help just holler.

If that were MS Windows 7 I would say:

SSD > 4-to-8 mem upgrade >>>> CPU upgrade.

I have no idea how bootcamp affects memory usage. On Windows 7, when I am heavily multi-tasking I see perhaps 3GB memory usage (I have 8GB). That leaves the rest available for system caching which helps a lot.

Also, my experience is with the OCZ Vertex SSD. I have no experience with the so called "crappy" SSDs (reviews say Intel X-25 > OCZ Vertex > the rest).


EDIT: I have a 120 GB SSD and I think that is plenty. Sure, more space is better but I would be careful to get a fast SSD rather than the biggest SSD.

Unfortunately Apple doesn't give any information on manufacturer or speed for the SSD drive upgrades. I'll see if I can find any info elsewhere on the web. How is speed usually measured on SSD drives? I don't remember seeing stats on them usually.

start here: AnandTech SSD Relapse. That is a good overview, then read any new reviews.

from what I hear, manufacturer installed SSDs are crap (performance wise). Can't you buy a 3rd party drive and install it yourself or have a service center install it? A crappy SSD may still be better than a typical laptop HD (not sure though).

That is a fantastic article, thank you. Installing a third-party SSD is certainly an option. I just wish Apple would give out the drive information so that I can compare it to the options on the market

if you do 30 minutes of google research I'd be surprised if you could not find out what SSD they ship.

They'll never tell you. They simply put out bids for each production run. I build mission critical workstations and servers for AV post production (which is far more demanding than anything PT3 can dish out) and the financial industry and deal will hardware issues on a regular basis. The performance gains using SSD are decent as HDs are a major source of throughput bottleneck but for PT3 you'll only notice such gains while doing massive imports with full housekeeping and then only +/- 10% timewise. I'm not trying to discourage SSD as I use several myself I just think you might run into problems when installing BD/W7. I have a customer that is running an SSD mac lap that I think also has BC - I'll talk to him when I get a chance.

Mass RAM and cpu will get the job done.

Hmm, a faster HD won't speed up the loading of data and graphs? My main concern with PT3 is that it takes forever to load the data when I open my filters to more than the past 2 weeks or so. If I clear all filters by mistake it will freeze up for several minutes trying to load all the players and data. I thought this was mainly an HD issue since it has to get it all off disk. Will increasing my memory from 4GB to 8GB help more than an SSD upgrade?

I believe the AnandTech article discusses which benchmarks are relevant to DB access. the verdict being that a good SSD significantly improves SQL performance.

How much of the PT3 lag is due to SQL performance and how much is PT3 software is anyone's guess. Perhaps a PT3 person can comment on this aspect.

What are your current computer specs?

Say for instance that a 7200rpm x 32MB cache HD can run a report in 5 minutes - a SSD *might* do it in 4.5 - 4.75. Huge difference - no.

RAM will have a larger effect. W7 + BC + background (and foreground) programs have huge overheard and will be using a large portion on RAM. Add PT3 and Postgres, poker client, browsers, skype, antivirus to the mix and RAM can be eaten up very quickly. Anything +4GB will be good.

It depends exactly what you're doing. For the HUD, PostgreSQL access time is the biggest bottleneck, but keeping your cache up to date will help a ton here (in the latest beta there's now a bar to tell you how up to date your cache is). We've been working on improving responsiveness and the latest betas should have improved things.

If you're running custom reports, especially without an up to date cache, access time again becomes the biggest bottleneck.

If you ever want to see how long queries are actually taking, you can feel free to enable logging in PT3 and look at the PokerTracker.log file - all query times are noted in the log.

I'm currently running on 4GB of RAM. I will most likely upgrade to 8GB when I buy my new machine, but it sounds like an HD upgrade won't have that much of an effect. I'll still probably get the 7200 rmp upgrade, but I guess the SSD option just isn't worth it given the huge price increase?

I run a full database maintenance every other day or so, and I don't think I've ever really had any issues with the HUD. It's mainly when I attempt to view more than the past 2 weeks of data that PT becomes unbearably slow in loading up the data and graphs. Clearing all filters is a guaranteed 5+ minute lockup.

SSD might to be worth the extra $$$.

Is your computer clean and free of malware and such?

What version of PT3 are you running? Did it start after you installed a recent beta? Sounds like you might be suffering from a few issues that a few of us have had since the recent betas. I understand they are working on them and should have most corrected on the next release. You might want to contact support and submit your logs from a freeze.

We use almost no CPU so memory and hard drive speed are always the best options.

As far as I know I only use it for playing poker and browsing 2p2, but being a native Mac user I'm probably not as smart with the antivirus stuff as I should be. I'll look into doing some scans.

I'm running the latest beta, but the problem has been there as long as I can remember. As soon as PT has to load more than 20k hands or so it takes forever to refresh. I figured it just had to do with the massive amount of SQL queries it had to run to reconstruct all the data on each refresh, so it's probably a combination of HD access speed and available memory. I'll have to keep an eye on active memory when it's being slow next time to see if that's maxing out.

I try to keep my filters down to the last day or so during my sessions so it's only really a problem when I want to do longer term reviews. It will definitely be worth spending the extra money on upgraded hardware to speed it up, I just want to make sure I'm upgrading the right things.

Awesome, thanks.

Make sure you update your cache; it should be much faster in Beta 30.6 . . .

define "forever". Presumably you can do a test and time the delay to the nearest second and report back saying something like:

I filtered "all hands for the last 2 months (37k hands) and it took 12 seconds before the graph appeared".

Kraada, if the cache is up to date, would IOMeter measurements be relevant at all? How about if all the hands being filtered are not cached? For non-cached, I guess I'm asking if any particular IOMeter test correlates well with SQL DB performance.

If the cache is up to date, the i/o hit should actually be pretty small. Really with an up to date cache you shouldn't have any performance problems regardless of database size. I've been doing testing on 9 tables with a HUD using close to 700 stats with two databases, one 500k hands and one 15.1M hands and I don't have any troubles (and my HD is a standard 7200 rpm drive, though I do have 8G of ram).

When you're using a filter on the main PT3 window the cache is disabled so all of the stats are grabbed and calculated from the database, which is why things will take a bit longer when you're using filters (though if you filter enough then the data you're looking at is small . . .).

AFAIK the SSD will improve HUD performance more than anything else. At least, the people behind HEM recommend it above all other upgrades.

As for the SQL-performance, remember to tweak your postgres settings. These will enchance your performance much more than a faster harddrive.