Performance issues – video card related?

[Bret Williams]

You’re comparing cost of iMac, with a $1000 screen (yes expensive, but the best out there) to and i5 with onboard graphics? It’s hard to spend 3k on an iMac without going for the somewhat unnecessary extras or upgrades. Did you buy the RAM from Apple? That’s 400 bucks down the drain compared to OWC or Amazon. Or did you get a 3TB fusion drive? Any fusion is kind of a convenience factor. But a decked out iMac with best 2gig graphics card they offer, and an i7 (not an i5, huge difference) will cost you around 2600 even with a 1TB fusion. Get the 32gigs of RAM from OWC and you’re at $2800. Lose the fusion and go with a 1TB (remember you should only put Apps and system on the internal) and you save another $250. I for one see no point in Apple Care either, but that’s me. I don’t buy extended warranties. They always make money for the person selling them. The only way to come out ahead is to never buy them. Kinda like gambling.

So If you upped your processor to an i7, and had a graphics card, wouldn’t that make it more expensive than the iMac? I still don’t get the whole hackintosh thing, although I’m still curious. Aren’t they also limited to accessing a single processor on the Mac side? So best you can do is a 6 core i7.

[Bret Williams]

Sorry Michael, I confused your specs with Eddie’s. I’d like to know your specs, and Eddie’s cost for his specs as well. This 1/3 the cost is intriguing because I just don’t see that.

[Eddie Adams]

It was about $900 for the whole tower. I already had the monitors and reference speakers. Thats without an added GPU. I can list the specs in more detail later on. What I don’t like about the imacs is anything you want to add is external. Hackintosh allows me the flexibility to add drives, DVD burner, coolers and fans, etc.

[Michael Garber]

Hey Brett-

No worries! My feeling on AppleCare is that increases the resale value of the Mac. It’s peace of mind for buyers who might not be in the industry. It casts a wider net. Also, I’ve had to use it a couple times in the past and it paid for itself. So, for me, it’s worth it.

Eddie’s system might be $800, which is great. But he’s on here asking why his system isn’t working that well. That’s time spent and frustration added. He’ll probably need to spend up to $800 for a new CPU and GFX Card. At most $500 for a new gfx card.

Hackintosh part prices below are current and possibly not reflective of what I paid. But it’s very close. This gets you mostly in parity with a top of the line iMac (minus Bluetooth and WiFi, which cost about $46 total for the components). Here are my specs:

Intel Core i7-3770K $320
Gigabyte GA-Z77X-UP5-TH (has 2 thunderbolt ports) $245
32GB 1600Mhz DDR3 $217
Corsair Carbide 500R $158
Corsair 650 Watt Modular $90
GeForce GTX 680 $497
480GB SSD Drive $363
Syba SATA III 6Gbps PCI-e Card $17
StarTech.com 3 Port 2b 1a 1394 PCI Express FireWire Card Adapter PEX1394B3 $63
Cooler Master Hyper 212 EVO: $33
I owned a Kona 3, which I installed and is working fine.

TOTAL: $2003

iMac System Specs:
3.4GHz Quad-Core Intel Core i7
8GB 1600MHz DDR3 SDRAM-2X4GB
1TB Fusion Drive
NVIDIAGeFrc GTX 680MX 2G GDDR5: $2600
AppleCare: $170
RAM from OWC: $270

TOTAL: $3040 (before tax)

Now, here’s the kicker. While Geekbench isn’t the be-all end-all, I think it’s a nice cross-reference for the technologically-barely-informed (such as myself ;-).

My iMac 27″ top o’ the line gets a healthy 14268
The Hack Pro gets a substantial 14124

For all intents and purposes, they are equal systems. One cost $3014 and comes with a warranty. The other cost $2003, works great, and is expandable, but has risks. Again, if you need to purchase a monitor, keyboard and mouse for the Hackintosh, that will add an additional amount and bring them much more in line price-wise.

Let’s look at the percentage difference if we take out AppleCare to get the systems more in line.

iMac: 2845
Hack Pro: 2003
Percent Difference: %30

Now, let’s look at potential resale value and upgrade costs. This is assuming a new iMac will cost about the same amount in 2 years:

iMac late 2012 resale value in 2 years (conservative): $1250
iMac 2014 (high end) potential cost: $ 2900
Out of pocket difference: -$1650
Time needed to reinstall software and test new system: 1 day (based on purchase of iMac)

Hackintosh upgrade option 1, resale and build from scratch:
Hackintosh Late 2012 Cost: $2003
Hackintosh Resale Value in 2 years: $400
Hackintosh Late 2014 potential cost: $2000
Out of pocket difference: -$1600
Time needed to research new system and find best prices on parts: countless nights on Hackintosh boards
Time spent building system: 1 day
Time spent testing parts and debugging: 2 days (conservative)

Hackintosh upgrade option 2, upgrade CPU, Mobo, and RAM
New Motherboard, Processor, RAM (assuming all current PCI Cards work properly and base prices stay the same): $782
Time spent researching whether all parts will work with new system and what CPU/Mobo to get: 2-3 evenings on hack boards
Time spent installing new parts: 2 hrs
Time spent reinstalling OS, debugging: 1 day (conservative)

So there you have it. At most, there’s an $870 difference (not including time spent) in favor of an equally spec’d Hack Pro and a top-end iMac over a 2-year period. Divide that out by 2 years and that’s $37 per month.

The ultimate question I propose (which I’ve already answered for myself): Is it worth it for you to build one?

Michael Garber
5th Wall – a post production company
Blog: GARBERSHOP

[Bret Williams]

Definitely not worth it. I could argue the hackintosh is more expensive in many ways. The built in 1k monitor is the big difference obviously. With much cheaper options, but none even close to as good. And apple resale value (my 2011 imac is still 2 worth around 2k) is a kicker. It’s nearing 2 years old and for 2 years use has cost me less than $500. I’ve sold many macs on eBay and was always amazed. iPhones too. I sold my bused iPhone 3GS for $350. WTF?

But it would be fun to build. My wife would kill me of course since we don’t have time for me to have fun with 2 small kids.

I used to think I needed a tower, but really, thunderbolt changed all that. I have a raid under the desk that is much smaller than a tower. So drives built into the machine are a non-argument.

When I bought my current system, it was a maxed out iMac with Pegasus raid for $4000. Plus $70 of 16gigs of RAM from Amazon. The iMac $2350. The pegasus $1600. No sales tax. No shipping. Everything from B&H. Maxed out. Same iMac is still that same price. But now sans a DVD burner. One doesn’t need the fusion. But the 1 gig fusion brings it up to $2600.

[Eddie Adams]

So I dropped in the EVGA GTX 660 GPU. Although I do see a small speed increase in how fast FCP X refreshes Waveforms in clips, it is still slow to load thumbnails – there was absolutely no increase in speed there. This leads me to believe that the issues is not in fact related to the GPU – that perhaps FCP X has issues with thumbnails and load times? Has anybody else experienced this? I’m running all footage off a 1 TB WD Black drive, so it’s a fast as it gets in that regard. Clips are 1080i 29.97 ProRes 422 – so I really dont see where all this lag could be coming from. Any ideas? Thanks.

[Michael Garber]

Have you done a speed test on the Caviar Black drive? You might be better served with a raid of 2 or more drives.

I could be wrong, but I believe that FCPX uses the CPU for thumbnail cache. You can also try deleting your thumbnail cache and starting over. Sometimes that can get corrupted.

But in general, even on a fast Apple-built system, FCPX tends to get sluggish with thumbs and waveforms. For instance, it is constantly redrawing waveforms in the timeline (although I have yet to test with .8 to see if it fixes any of that). This constant redraw can be a headache when editing.

Michael Garber
5th Wall – a post production company
Blog: GARBERSHOP

[Eddie Adams]

I have two main drives that I use in the system:
The first drive (my Edit Drive) is a RAID 1 mirror from two WD Black 1TB drives.
The second drive is a 7200rpm Toshiba.

Speed tests show (copying a 8GB file across the two drives):

Write speeds to both WD Black drive(s) were around 115-120 MB/s (same read speeds from Toshiba)
Read speeds from the WD Black drive(s) were around 60 MB/s each drive (120 MB/s write to Toshiba)

What’s interesting is that while the RAID 1 drives seem to be WRITING as a mirrored pair (both at the same speeds of 120 MB/s), they seem to be READING from the drives as a striped pair, splitting the workload between the two drives. Am I correct in seeing it this way?

If this is all correct, shouldn’t FCP be reading from the drives as if they were a RAID striped pair, thus allowing maximum read speeds?

I’d be interested to learn if FCP X is in fact using the CPU to redraw waveforms and thumbnails…

[Michael Garber]

You don’t want to edit from a mirrored drive set. That will slow you down. Best to edit from either a RAID 0 or a RAID 5.

Michael Garber
5th Wall – a post production company
Blog: GARBERSHOP

[Dave Gage]

[Michael Garber] “FCPX tends to get sluggish with thumbs and waveforms. For instance, it is constantly redrawing waveforms in the timeline (although I have yet to test with .8 to see if it fixes any of that). This constant redraw can be a headache when editing.”

I agree. Unless I need to see the audio waveforms, I shut down the timeline thumbnail view with the little light switch hud. As someone suggested here, I’m experimenting with also closing the Inspector when not in use also. But, certainly shutting down the waveform re-draws makes a big difference, at least for me.

Dave