How much more expensive will the next Mac Pro be?

[sgunes]

what about non overclocked?

Let's figure it out (assuming you are not doing encryption):
980x: 6 x 3.33 GHz = 19.98 GHz + turbo boost.

Octocore MP: 8 x 2.33 GHz = 18.08 GHz.
Octocore MP: 8 x 2.66 GHz = 21.28 GHz.

So, yes the unoverclocked Gulftown will beat any day the Octocore 2.26 GHz.
It may be too close to call against the 2.66 GHz Xeons.

Overclocked:
980x: 6 x 4 GHz = 24 GHz or even
980x: 6 x 4.3 GHz = 25.8 GHz

Octocore MP: 8 x 2.93 GHz = 23.44 GHz

The only Apple MP model that can beat the overclocked Gulftown would be the dual six-core Xeons.
This does not even take into account that you can run your memory faster than Apple usually does.

 

[Kpjoslee]

Let's figure it out (assuming you are not doing encryption):
980x: 6 x 3.33 GHz = 19.98 GHz + turbo boost.

Octocore MP: 8 x 2.33 GHz = 18.08 GHz.
Octocore MP: 8 x 2.66 GHz = 21.28 GHz.

So, yes the unoverclocked Gulftown will beat any day the Octocore 2.26 GHz.
It may be too close to call against the 2.66 GHz Xeons.

Overclocked:
980x: 6 x 4 GHz = 24 GHz or even
980x: 6 x 4.3 GHz = 25.8 GHz

Octocore MP: 8 x 2.93 GHz = 23.44 GHz

The only Apple MP model that can beat the overclocked Gulftown would be the dual six-core Xeons.
This does not even take into account that you can run your memory faster than Apple usually does.

No.....You cannot simply add clockspeeds together and assume which one will outperform another. It really depends on which application you are going to be using.

 

[Gonk42]

Bench marks for overclocked 980X

For those interested in overclocked 6-cores, the latest edition of Custom PC reports on their build where they overclocked it to 5GHz. This will probably appear on bit-tech.net some time but I don't think it is up yet.

The Cinebench score was 39,391 (R10)

Personally I use a workstation for stability rather than outright speed, so would avoid overclocking - though this is probably too conservative as a well cooled overclocked chip is probably as stable as a less well cooled chip running at stock frequency.

One advantage a 2x4 core arrangement may have over a 1x6core (apart from the extra two cores) is that it has 6 memory channels vs 3 on the single chip setup.

 

[alphaod]

Why no 4-socket Nehalem-EX system with a total of 32 cores (64 if you count Hyper-Threading)? The sky is the limit.

Because not everyone wants a cooling tower to cool their desktops

 

[sgunes]

No.....You cannot simply add clockspeeds together and assume which one will outperform another. It really depends on which application you are going to be using.

Actually you can.
In my examples the higher clocked 6-core will beat the lower clocked 8-core as it is easier to make use of fewer higher clocked cores than more lower clocked cores (given the same microarchitecture and memory speed) . The few exceptions are Cinebench R10 and other programs that will use every core and thread.

 

[Hmac]

I think prices on equivalently-positioned machines will go up a couple hundred $$. Apple is moving toward a consumer and mobile sales paradigm. It doesn't appear that they have any intention of trying to penetrate the increasingly rarified atmosphere of high-powered desktops, so I wouldn't look for any sweet deals on Mac Pros or big ACD's.

 

[Max(IT)]

I'm fairly confident that an over clocked 980x can outperform an 09 2.26 8 core Mac pro

Wtf ... how can you keep talking about overclock when speaking of professional machines ?!? That's well beyond me ...

 

[nanofrog]

Wtf ... how can you keep talking about overclock when speaking of professional machines ?!? That's well beyond me ...

Perhaps it's that gamers used to OC'ing thier rigs are entering the professional market (i.e. recent graduates), and as independents, see OC'ing these systems as a way to eek out improved performance without the understanding that reliability is even more important than performance.

 

[Kpjoslee]

Actually you can.
In my examples the higher clocked 6-core will beat the lower clocked 8-core as it is easier to make use of fewer higher clocked cores than more lower clocked cores (given the same microarchitecture and memory speed) . The few exceptions are Cinebench R10 and other programs that will use every core and thread.

In your examples, 2.8 dual core will always have to be twice as fast as 2.8 single core, but that is not the case is it? Also, adding clockspeeds together is COMPLETELY WRONG way to estimate performance. 6 core vs 8 core is far more complex to assume as it heavily depends on how the application is able to utilize multiple threads.

 

[wisty]

In your examples, 2.8 dual core will always have to be twice as fast as 2.8 single core, but that is not the case is it? Also, adding clockspeeds together is COMPLETELY WRONG way to estimate performance. 6 core vs 8 core is far more complex to assume as it heavily depends on how the application is able to utilize multiple threads.

It's like saying your Ferrari can out-pull a team of draft horses. It's wrong, but conservatively wrong.

It's ignoring the reliability of EEC RAM though.

 

[sgunes]

In your examples, 2.8 dual core will always have to be twice as fast as 2.8 single core, but that is not the case is it? Also, adding clockspeeds together is COMPLETELY WRONG way to estimate performance. 6 core vs 8 core is far more complex to assume as it heavily depends on how the application is able to utilize multiple threads.

If you re-read my post you see that I mentioned programs like Cinebench that will use every core and I mentioned that you have to have the same microarchitecture, memory speed and other factors identical to make it a fair comparison. We had this discussion here multiple times before regarding more GHz versus more cores.
In general you are better off (once you are over 2 cores) to have fewer faster cores than more slower cores. So a 4 core 4GHz machine will be faster than an 8 core 2GHz machine except for pure rendering when they maybe equal.
My point was that an overclocked Gulftown hackintosh will only be beat by the fastest 12-core MP 2010. Until we have all in hand and can benchmark them it is an assumption. What we have right now are Cinebench scores for overclocked and regular Gulftown.

 

[nanofrog]

If you re-read my post you see that I mentioned programs like Cinebench that will use every core and I mentioned that you have to have the same microarchitecture, memory speed and other factors identical to make it a fair comparison. We had this discussion here multiple times before regarding more GHz versus more cores.
In general you are better off (once you are over 2 cores) to have fewer faster cores than more slower cores. So a 4 core 4GHz machine will be faster than an 8 core 2GHz machine except for pure rendering when they maybe equal.
My point was that an overclocked Gulftown hackintosh will only be beat by the fastest 12-core MP 2010. Until we have all in hand and can benchmark them it is an assumption. What we have right now are Cinebench scores for overclocked and regular Gulftown.

But you're talking about a benchmark program, which isn't real world results. Most software can't take advantage of all the cores, even if it is multi-threaded. There's a little that can, and rendering tends to be where you find it (for common usages by MR members).

There's other applications that can as well, but it's usually in server or scientific areas for workstation use (typically for simulations). In general, it's rare though, as software always lags behind the hardware.

Just look at Adobe's offerings. Multi-threading usually means 2 cores currently. Now if you can multi-task, you can run multiple instances at the same time, and use the system's power, but it's hard to do as I understand it.

 

[Max(IT)]

Perhaps it's that gamers used to OC'ing thier rigs are entering the professional market (i.e. recent graduates), and as independents, see OC'ing these systems as a way to eek out improved performance without the understanding that reliability is even more important than performance.

you could be right

 

[RubbishBBspeed]

Hopefully not much more, I'm already getting stick for it taking so long, if it arrives with a price increase then I think I'll be asked to resign (not joking). But if it doesn't come by the end of June then it wouldn't matter I won't getting one.

 

[Wild-Bill]

I hope they release the damn thing soon.

I probably won't buy it, but I'm interested in getting my hands on an HD5870 for my Early 2008........ that is, if they offer the card. Knowing Apple, the video card options will probably suck.

 

[Techhie]

I hope they release the damn thing soon.

I probably won't buy it, but I'm interested in getting my hands on an HD5870 for my Early 2008........

I will probably look for a good price on an '08 or '09 octad with AppleCare, so I can spend the rest of the money on more important things (like food other than ramen )