8 cores? are you kidding?

[revenuee]

when i stopped doing photo/video stuff i found a lot of things overkill

I started doing a lot more math operations on my computer -- particularly back testing strategies --- having a little more power is nice when you're running a simulation against 20 years of currency information

 

[FF_productions]

I wasn't saying that a faster CPU is all that's needed to convert an entire movie in a couple minutes. All I saying is that there really isn't such a thing as "over kill" for a computer. Things need to be invented before they get cheaper for everyone (not that the MacPro is that expensive compared to the competition anyways).

I'm just talking in general. Can't the Mac Pro's get 20+Gig IO? If I'm wrong please tell how much it really is.

 

[neverownedapc]

i didn't mean to upset anyone who has an actual need to have the best since they actually utilize it for extremely large programs that they use alot or that need that kind of power, so if i came off wrong, sorry. i'm just saying that the market is really out of control just because people outside of that realm have this "need" to have something and they don't even know they don't need it like the soccer mom with the hemi under the hood. it's kind of funny.

 

[Lord Zedd]

More CPUs is always better, even for internet browsers. GHz is a huge limit right now, we won't be seeing 4GHz+ for several years but we will be seeing 6, 8 and 10 core CPUs by then.

Even browsing the forum on one monitor and watching a movie on the other I'm using most of two cores. It would be nice to have 3 or 4 cores instead of just 2 for the extra available power.

 

[Cromulent]

Let me say it more clearly.

All those cpus are a waste if you can't get data to them. Which you can't with this architecture around them. You aren't investing in a machine that will "last you longer" because the real bottlenecks are in the IO not in cpu power. In 3 years buses will be faster, and (hopefully) drives will be as well. Therefore, an 8 core system isn't going to "last longer" than a quad because you need the faster IO that the current system doesn't have to take advantage of the cpus.

That isn't true at all. Some mathematical functions are very light on data but require immense processing power.

Anyway if you are constrained by IO to such a large extent then just invest in an optical RAID 5 array with 5 - 10 drives in it and be done with it.

 

[saltyzoo]

That isn't true at all. Some mathematical functions are very light on data but require immense processing power.

Anyway if you are constrained by IO to such a large extent then just invest in an optical RAID 5 array with 5 - 10 drives in it and be done with it.

Yeah, it is true at all. I did say "unless you are doing certain things" which nobody here is doing.....

And a RAID 5 isn't going to get the data over the bus to the cpus. IO is a HUGE bottleneck even with 2 cores. 4 cores is ok, 8 is ridiculous.

 

[Karpfish]

If they have the money, is it really an issue that they want one?

 

[hexonxonx]

If they have the money, is it really an issue that they want one?

Shouldn't be to anyone. I got Mac Pro for two reasons:

4 TB storage plus tons of external possibilities
The length of time it takes to rip DVDs.

I have many box sets of DVDs I want to rip and gave up on them when my iMac would take too much time. Now I can rip a whole DVD in 15-20 minutes.

The above reasons most likely seem crazy to allot of people but to me, buying the Mac Pro was extremely well worth it.

I could have easily have gotten a 4 core to do the same but the 8 core was readily available at the Apple store si it's what I bought. I don't like waiting for BTO.

Yes, I am enjoying and getting my moneys worth on my Mac Pro.

 

[OlBlueHair]

Using this logic, everybody should drive a Hyundai Accent. Nobody needs to go 150 miles an hour to drive to the post office and the video store.

(post from a guy who drives a ****** Accent but has a new top of the line Mac Pro)

 

[CWallace]

While I/O performance is an issue, the Mac Pro benefits from using workstation/server parts which helps shovel data around the system quicker.

Our workstations with two dual-core Xeon 5150s run rings around our desktops with a single quad-core Core 2 Quad Q6700 when it comes to network bandwith (both have Gigabit Ethernet connected directly to a Cisco gigabit switch) and HDD performance (both use SATA 3.0) thanks to the more robust chipset and systemboard architecture used in the workstations.

 

[Decrepit]

And yet it's all still sitting still waiting for that 7200 rpm disk and slow ass system bus.

Unless you're breaking military grade cyphers or searching for ET, that cpu power is a complete waste if you can't get data in and out of it any faster.

Throwing in more cores is cheap, faster IO is a pain in the butt and requires a lot more expense to build. It doesn't matter that you can't use it, people think "bigger" has to be better so they'll buy it. PT Barnum knew the American consumer well.

Bus speeds will be overcome. We had 33 MHz, and now 1066MHz front sides.

Disks are available at 15000 rpm

Memory speed is increasing all the time, and access to 64bit addressing will mean more systems that can run exclusively in memory instead of using the disk at all.

When the real memory that can maintain its state during a non-power situation comes down (way down) in price, we're going to see some neat stuff. Moving parts are bad news. We're going to be able to have extremely thin systems that are silent and unbelievably fast.

I think a lot of technology sat on the sideline, waiting for 64-bit. With multi-core 64-bit systems becoming more mainstream, we should see a push to finally rectify the disk and bus speed issues. Before there was no need. Now we can produce applications and operating systems that can take advantage of it. Vista seems slow on the highest end stuff. OS X seems to scale nicely from the little bit of it that I've seen on high end boxes.

 

[zmttoxics]

Bus speeds will be overcome. We had 33 MHz, and now 1066MHz front sides.

Disks are available at 15000 rpm

Memory speed is increasing all the time, and access to 64bit addressing will mean more systems that can run exclusively in memory instead of using the disk at all.

When the real memory that can maintain its state during a non-power situation comes down (way down) in price, we're going to see some neat stuff. Moving parts are bad news. We're going to be able to have extremely thin systems that are silent and unbelievably fast.

I think a lot of technology sat on the sideline, waiting for 64-bit. With multi-core 64-bit systems becoming more mainstream, we should see a push to finally rectify the disk and bus speed issues. Before there was no need. Now we can produce applications and operating systems that can take advantage of it. Vista seems slow on the highest end stuff. OS X seems to scale nicely from the little bit of it that I've seen on high end boxes.

The 8 core mac pro actually has a 1600mhrz front side bus. I wouldn't call that "slow ass" at all, in fact I believe its the fastest at the moment. Isnt it?

 

[saltyzoo]

The 8 core mac pro actually has a 1600mhrz front side bus. I wouldn't call that "slow ass" at all, in fact I believe its the fastest at the moment. Isnt it?

Actually, you're right, that is pretty fast, I wasn't aware they bumped it up the last update. It's up to quad core performance now.

 

[Cromulent]

And a RAID 5 isn't going to get the data over the bus to the cpus. IO is a HUGE bottleneck even with 2 cores. 4 cores is ok, 8 is ridiculous.

Hard drives are going to bottleneck waaay before the system bus does. So yes a RAID 5 will help.

Edit:

From HP:

The 1600 MHz Front Side Bus (FSB) is a quad-pumped bus running off a 400 MHz system clock making 12.80 GBytes per second data transfer rates possible.

Source : http://h10010.www1.hp.com/wwpc/us/e...293-337097-64293-3642734-3642735-3642737.html

 

[Firefly2002]

Here's some of GS320 specs BTW, keep in mind this is an obsolete machine that's not sold any more!

# Up to 32 1.224 GHz Alpha processors
# Up to 256 GB of ECC memory
# Over 51 GB/s aggregate internal bandwidth
# 16 MB ECC memory onboard cache per CPU
# Up to 224 PCI slots on 64 PCI buses

The newer GS1280 model has over 200 GB/s of I/O bandwidth, 512 GB memory, and 64 CPU!

Wait, they still make Alphas? As in DEC?

Lol, I remember being like 8 or 9 or something and looking in a magazine and seeing a 500 MHz DEC Alpha and being like WTF? The closest competitor was I think Apple at 200 MHz.

But yeah. Does DEC still make em or is that HP you linked..? Either way the system sounds fast

 

[Gloor]

Try doing CG or any 3D and try the difference when it comes to rendering. Try 2core and 8 cores and you will see the huge time save. Then you can come back and talk nonsense here. Thanks

 

[saltyzoo]

Try doing CG or any 3D and try the difference when it comes to rendering. Try 2core and 8 cores and you will see the huge time save. Then you can come back and talk nonsense here. Thanks

Not as much of a difference as you would with better IO, faster drives, and a boatload of RAM and a good quad core, but yeah.

 

[Cromulent]

Not as much of a difference as you would with better IO, faster drives, and a boatload of RAM and a good quad core, but yeah.

Where might I ask is your experience coming from when you post things like this? Or are you just saying what you think is true? Going from 4 core to 8 core is a huge upgrade in terms of 3D rendering and because 3D is completely CPU bound, adding more RAM or faster hard drives is going to do exactly sod all to the speed of rendering.

 

[Tbarr]

Wait a minute.

The Mac Pro has 6 (SIX) SATA buses.
A simple RAID 0 using 2 (TWO) of the 6 (SIX) Buses that use the newer Terabyte drives has a throughput of 200MB per second.

The memory bus has a 12.80 GB per second throughput, so forget about that.

Now, what movie are we talking about? A normal dual-layer DVD has around 8GB of video. Let's see...
200MB per second is around 5 seconds for 1GB.
5 seconds X 8GB is 40 SECONDS. I believe that's less than 2 (TWO) Minutes.

What about standard definition DV? Well, 1 (ONE) hour of SDDV is around 13GBs. Now, 5 seconds X 13GB is 65 SECONDS. Still under 2 (TWO) Minutes.

2 Hours of SDDV? OK... Two hours of DV would take TWO minutes and 10 seconds.
Hmmm, maybe use 3 (three) Terabyte drives RAIDED to get 300MB per second? That would still leave THREE (3) SATA buses for three more drives. Yes, there are kits that will allow your Mac Pro to mount two SATA drives in the lower Optical Bay, allowing 6 (SIX) SATA Internal drives.

They call it a Mac Pro '64 bit Workstation' for a reason. If YOU don't need a 'Workstation', then buy something else. I like mine.

 

[Gloor]

Not as much of a difference as you would with better IO, faster drives, and a boatload of RAM and a good quad core, but yeah.

Assuming something is not the same as knowing something. Try it and then come back.

Cromulent said it correctly. That is what I meant and that is how it is. Rendering is CPU only. Each core => time saved.

 

[saltyzoo]

Assuming something is not the same as knowing something. Try it and then come back.

Cromulent said it correctly. That is what I meant and that is how it is. Rendering is CPU only. Each core => time saved.

The only thing that is cpu only is a no-op.

 

[Cromulent]

The only thing that is cpu only is a no-op.

I didn't say CPU only. I said CPU bound. Which means that the CPU is the component which limits the speed of the operation, not the I/O, not the RAM, not the hard drives. Just the CPU.

Getting faster RAM or hard drives will do very little, if anything to increase the speed of rendering.

 

[saltyzoo]

Maybe this will clarify

Rendering is CPU only.

The only thing that is cpu only is a no-op.

 

[Tbarr]

Mac Pro Performance

http://www.apple.com/macpro/performance.html

Application Performance:

Final Cut Studio HDV encoding
Up to 2.3x greater with standard 8-core system

Photoshop CS3 45-filter function
Up to 1.3x faster with standard 8-core system

Maya 2008 Mental Ray 3D rendering
Up to 2.1x faster with standard 8-core system

Logic Pro 8.0.1 Number of concurrently playing Platinum Verb reverb plug-ins
Up to 2.3x greater with standard 8-core system

PyMOL 1.0 Wizard benchmark CPU-based tests
Up to 1.4x faster with standard 8-core system

Mathematica 6.0.1 Personal Grid Edition MathematicaMark6
Up to 1.8x faster with standard 8-core system

CINT2006 Rate and CFP2006 Rate. Integer and floating-point calculation results
Up to 1.9x faster with standard 8-core system

Don't tell me... Since it's on Apple's site, you don't believe it.

 

[Gloor]

CPU only and CPU bound is almost the same. In other words, CPU does the work when it comes to rendering. That's what I meant so don't catch the words but catch the meaning. (I'm not native english speaker)