You won't get a lot (any?) benefit from RAIDing as the benchmark is designed to stress graphics cards. Very little writing is done,. I recall we did that and it made no difference with real hard disks as opposed to SSDs,
I've never got the hotel screen to work, I have tried for a few years, it always crashes with my 280X cards.
I gave up on it.
I tried my dual 280X cards in various PCI slots on my Z97X MB, I forget the speeds but some are twice the speed of the others. No noticeable difference to me. I still did this benchmark in 11-13 secs (from memory).
I also benchmarked 5770's in an old Mac Pro with the slow and very slow PCI slots and it made no difference.
I thought that the 290X cards are still problematic. e.g. dual.triple displays are a pain and need a certain combination of screens on certain ports and restarting in certain ways. Also some 290X cards don't quite work, choppy graphics, black screens etc.
The 280x cards are pretty well supported now and work very well. Mine do anyway
I'd advise against the 290X until it becomes as simple and reliable as the 280X. Of course if somebody can point to a guide that shows the 290X working easily and flawlessly my comments are withdrawn.
I looked hard when I brought my graphics cards as to what were the most reliable. Certainly 8-10 weeks ago the 290X were not reliable enough (IMHO) to be used as mainstream cards.
I would also be wary of using dual 290X or 280X cards in a genuine mac pro. There has been a massive amount of nasty discussion about whether a genuine mac can support these heavy duty cards. Whilst the PSU can easily drive it, the connectors and apparently the PCB track layers may not be designed for the max draw of 250W at whatever voltage and amperage is needed. I have no expertise to decide for myself, but I wanted dual graphics cards in my Mac Pro and nobody could tell me for certain if it was OK. I know dual 5770 cards work in the Mac pro as I did that myself, but the 280X and 290X cards draw twice as much power per slot.
I would advice caution as its an expensive mistake (if it is one) to make. Dual 290X cards in a Hackintosh is perfectly OK though Weird, a hackintosh can make a better Mac than Apple.
I wasn't implying you were fudging the numbers. I wondered if the rendering had been pre-done.
The only other benchmark I know of with a Mac Pro was around 19 secs and that was with dual D700 cards which I thought would be quicker.
I have no idea what the memory of the other Mac Pro was but if you've managed to get the tests down to 15 secs thats great.
Are you sure about those timings on the MacPro?
Your timings with a single D500 card is a whole four seconds quicker than a new Mac Pro with twin D700 cards.
I asked this question some time ago and didn't get much of a response apart from probably not, but no one could say 100%, however I tend to agree with them. Offloading rendering to two different architectures sounds awkward and difficult.
I have virtually the same set up as yourself, the differences are the CPU is a 4790K running at 4.0Ghz, more memory (32GB) and different 280x boards, I have 2 xSapphire Vapour X 3GB version. I also have the Z97X board as opposed to the Z87X. I do actually have an identical Z87X but thats in the process of becoming an ESXI server so its difficult to run it again.
I can't get the Bruce X benchmark down to 18 secs unless I pre-render it. Without the pre rendering, its around 24 secs, with the pre rendering its around 12-13 secs.
Are you pre-rendering or bringing the Bruce XML file and doing the export without background rendering in advance?
If you're bringing the XML file in and just getting the output file, thats a massive difference in speed to my system which is for all intents and purposes the same.
Second R 280x arrived this morning. After faffing around and wondering why it didn't work, it turned out a USB card in a different slot messed things up. Absolutely no idea why
Anyway, did the Bruce X benchmark again and it rendered in 24 secs.
Gigabyte Z97X UD5H
i490K 4.0Ghz Not over clocked as I want stability.
Sapphire R280X Vapor-X 3GB OC. Apparently these overlock but haven't tried that yet.
Installation was pretty simple and I never used the onboard graphics card at all, went straight to the 280X.
Noise levels seem OK to say there are six fans on the graphics cards. To be honest I can't really hear it even with the sides of the case off.
Now off to try them in Crossfire mode playing games
Look forward to seeing the results from your two 280x.
I am just building a new hack with a Z97X board and an i4790K 4.0Ghz CPU. I have one 280X installed but am getting some issues with the machine simply rebooting with no warning or logs. I suspect some hardware issues but am still tracking it down.
My second 280X arrives Monday (I hope).
With a bit of luck I'll have worked out the hardware issue and will post new benchmarks for this test.
One of my dual 5770 cards in my Hackintosh needs replacing. I have dual 5770's that drives three monitors.
Whilst I can simply replace the 5770 with another 5770, I wondered if I should take the opportunity to upgrade the failing cards and thought about a 7970 or similar.
I was wondering if anybody had ever run a different AMD card with a 5770. I know that the nVidia cards and the AMD cards do not play well together, benchmarked that setup and laughed but I can't seem to find anybody who has run multiple AMD cards and FCP X.
The reason for not moving straight to a single 7970 is that I like the three monitors and I can't go to dual 7970's as my PSU won't support the power drawer. Additional PSU plus dual 7970's is expensive
So, anybody had any experience in say a 5770 AND a 7970? I've no doubt it will work, but will it work effectively?
Thanks for reading.
I tested with and without monitors connected and found zero difference in speed to be honest.
The 5770 does come in single slot models, I'm 99% sure I have one but am away at moment so can't check. Asus comes to mind though. I will be back tonight and can check then.
Can't comment on reflashing the 280X, all I would say is to check very carefully which ones are known to work as opposed to might work. The 6-8 pin adaptor is available, though a little obscure to find. I forget who I got mine off but suspect it was fleaBay.
Adding a second 5770 is pretty easy as well as cheap. You can use an off the shelf 5770 so you don't need to go Apple branded. I found a second 5770 moved my benchmarks down into the 31-32 second mark, though I have a newer 4770 intel chip.
You *may* need to be careful on the higher specced cards as they draw more current. There is some debate on whether the genuine Mac Pro's can deliver the amount of current needed by the higher specced cards. I followed a thread somewhere on what cards could get installed in what slots with whatever power draw they had. It got very technical, very, very emotive and in the end a massive slagging off took place on both sides.
Its confirmed that the 7970 works on a later Mac Pro
All I would suggest is that you see if anybody else has the 7970 working on the same model as you have.
The R9 280x is really a 7970 brought up to date. I believe it consumes less power BUT you might not get the Apple screen on boot up. Other forums may be able to help on the suitability.
Personally speaking, I think the 5770 cards are excellent value. I ran two of them in my Mac Pro and got excellent performance and stability. I still have the cards but in a non-mac unit and they still run very well. I'm struggling to justify any upgrade to 7970's or 280x. My heart says upgrade, my wallet says no.
Any information you have on the GT120 or of you do decide to upgrade is always welcomed.
Thanks for the reply. I was figuring that your answer was the likeliest one but was hoping I was wrong
I know people run the older 8800 and other cards graphics cards together, so was pretty certain it would work but suspected it would be sub-optimal.
I was aware that the 7970 was pretty much identical to the R280x but I tend to use the older nomenclature as I'm more familiar with that than all this new fangled Rxxx stuff
My logic is that if I reconfigured the system to have 1 x 5770, 1 x R280x, and if FCP X used the R280x then thats fairly similar to 2 x 5770's (AFAIK). This would mean that FCP X performance would be the similarish to what I have now if and only if FCP X prioritised the R280x over the 5770. If FCP X favoured the 5770 over the R280x then my performance goes down.
So the best I can get is the same performance (more or less), the worst I can get is down to single 5770 levels. Mmmm.... I think I've just talked myself out of an upgrade and saved me money.
Thanks for the information, I think my possible gains are not worth it for the possible downsides.
All the best,