core count = future proof?

[Revolution 11]

Well, that is a lot of upgrades, you can't justify those gains only to a CPU boost. I have upped the RAM on my laptop and more RAM makes this machine soar in response time and application "zip". Even my old Pentium 3 PC is almost always constrained in RAM, adding a gigabyte of RAM helps a lot in reducing drive thrashing.

 

[Hacp]

Never buy for the future. Always buy for the present. Right now, I'm using an Athlon II X2 250 OC'd to 4 ghz. Does everything I need. If I'm compelled to in the future, I'll upgrade the chip to an 8 core Bulldozer or Piledriver when I can get them for cheap used.

 

[Spjut]

You think a high core count for the next-gen consoles can help in making more games benefit from more than 4 cores?

 

[blastingcap]

You think a high core count for the next-gen consoles can help in making more games benefit from more than 4 cores?

Can't hurt, might help!

 

[OVerLoRDI]

Like others have said, a dual core vs single core the single core would get obsoleted so much faster. However beyond that, not so much.

I have a 1.8ghz core 2 duo in my laptop, I don't feel any need for more CPU in day to day tasks. 2 cores of pretty much any modern architecture can last you for ages in day to day tasks. Only exception might be the original Pentium Ds and to a lesser extent the socket 939 x2s.

 

[BD231]

As a programmer I feel I have some credibility to weigh in here...the flaw is not in the technology IMO. Rather, the flaw is in the economic incentives in the programming industry itself when it comes to software.

On the desktop can you name one program other than file compression (winzip, winrar, 7zip, etc) that is actually competing in terms of performance for your dollars?

How many programmers out there are facing a situation where they either invest in developing seriously multithreaded apps or risk facing losing customers?

I can't think of any, not in the mainstream market segment, outside of file compression and file compression is the epitome of a niche software application.

So why would any software company being managed by decision makers (not programmers) decide to increase the budget and spend more money on programmers to develop even better multithreaded software? Where is the ROI in that?

People who buy Adobe photoshop are going to buy it regardless whether it is fully multithreaded or just partially multithreaded. So the economic incentive for Adobe to fund the project of creating the next version of photoshop is basically to target a slight token improvement in multithreaded performance and push it out the door because they know people will buy it regardless the performance.

Even programs that are heavily multithreaded like povray and TMPGenc you are looking at customers basing their purchasing decisions not on performance but on feature set and capability.

I buy TMPGEnc not for its multithreaded performance but for its ease of use and uncompromised capability in extracting very good image quality at very low bitrates. I don't know a single person who prioritizes transcoding speed over transcoding quality when it comes time to open their wallet and buy the software.

And that, in a nutshell, is why I beleive the situation with multithreaded apps in the consumer space is what it is. Consumers don't base software purchasing decisions on the performance of that software, they base their software purchasing decisions on the capability of that software.

Even in gaming, I doubt you will find a single gamer who wanted to buy BF3 but chose not to because BF3 wasn't multithreaded enough. If they wanted BF3 then they bought it (or stole it, but that's a different discussion) and if they didn't get the FPS they want then they upgraded the hardware until they were happy with the final result.

So if you are a software project manager what are you going to do when it comes to making decisions on how to invest your limited project development dollars? Are you going to tell your programmers to go after better multithreaded performance or are you going to tell them to make sure they include feature XYZ, regardless how slow or fast that feature ends up being, so that the customer decides to buy the software?

The problem with TMPGen is the same of mods or harlots, you might loose some image quality but they'll still charge you for it. Sadly it's because of people like you (programers) we have to settle for crap software to make our hardware function properly. Blame the companies all you want ya lazy finger pointin programer .

 

[SammichPG]

I mostly agree. It's difficult to make many kinds of programs highly multithreaded.

Nevertheless, 2 cores with HT would be the minimum I'd consider if buying a new CPU for basic multitasking; heck even many games can at the bare minimum shove their audio and other threads onto a second core. IMHO, more than 4 cores is a waste for most consumers, though. Even the really parallel applications can often be run overnight or something.



Do browsers count even though they are mostly/all free? Because I appreciate how modern browsers are more multithreaded now than before, even if they are mostly/all free these days. And the devs get paid indirectly via the platform and ad revenue and such so there is an incentive to boost web browser market share.

Those applications could use a common standard for network clustering, I own 4 computers with a total of 11 cores (one tri core, one quad and two dual cores on laptop), some stuff like video encoding is not that intensive I/O wise and could work well with a 1gbit lan.
The performance I could get out of all my computers in a very parallelized task would be nothing to scoff at and yet I'm limited to the 4 cores of my q9550 unless I'm willing to dabble around linux or use professional software with network renderers.

Nodoby in the consumer space has any interest in it.