Can AMD salvage QFX with an in-house chipset?

Page 22 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Since I am going by what AMD has said, you are stating opinion.
You are stating AMD's marketing, not facts.

C2D was NOT released to defeat or even compete directly with X2, it was released to improve upon NetBurst.
The market doesn't agree. The market sees C2D competing with X2.

And no you don't need a better PSU. The average Opteron PSU will do just fine and those are usually between 600W and 700W (my current PSU is 500W). This is further exacerbated by SLI
Oh, yes it does need a better PSU. With the same components, QFX will need 130W+ more than an equivalent C2Q system. More power consumption is more power consumption... there is no free energy source in QFX systems

WHICH I WON'T HAVE.

Anand showed that full load on 8800GTX/RAID/FX74 was 456W. That's higher than C2Q but not cause for rewiring the house. The power it gives me with Visual Studio and SQL Server/Exchange/BizTalk for development makes it worth it.
Ever considered C2Q? Faster and cooler!

Everytime you say the same nonsense about two chips taking more power than one, it makes me want it more.
Well, I never said that...[/b]
Click to expand...


Read my sig. I will still be getting it.
 
it's real simple. jack is saying that engrs shouldn't have to optimize for ccNUMA, but it's OK that they optimize for SSE.

Without SSE Intel will still have few wins over AMD. It could be considered as much of a crutch as optimizing "first touch" routines to maintain coherency in RAM requests.
Click to expand...

Baron, I said no such thing --- please quote my post where I made such a statement.

What I, Wr, Pippero and others have tried, and failed, to do is educate you on what NUMA really entails and how that is dragging down the performance on the desktop. We have tried in vain to explain to you that the chipset is not the feature the drives the architectural platform to a NUMA memory model, the chipset is not responsible for scheduling execution on the CPU cores, and the chipset is not responsible for desginating the context of the thread and which node it should run through.

So at that I agree with the other posters, this thread should now be locked. It has bore no fruit.

Jack

You're not above me. You're just an anonymous voice on a public forum. I understand completely how the kernel and AMDs NUMA implementation work. That's why I'm confident that profiling can work along with "adaptive swapping." I'm sorry hat only regurgitation is available to you, but I am an engr. Teh type of person that makes the seemingly impossible possible.


And yes you did imply that no one would optimize for such a small-niche platform. I will not scan through the pages to find it.
If engrs felt like you there would be no Core 2. Nor any software optimization for any particular thing.

Give it up. It's a great idea and shows great promise for the multi-threaded future.

ah heck off chav.

To quote the British among us.
 
it's real simple. jack is saying that engrs shouldn't have to optimize for ccNUMA, but it's OK that they optimize for SSE.

Without SSE Intel will still have few wins over AMD. It could be considered as much of a crutch as optimizing "first touch" routines to maintain coherency in RAM requests.
Click to expand...

Baron, I said no such thing --- please quote my post where I made such a statement.

What I, Wr, Pippero and others have tried, and failed, to do is educate you on what NUMA really entails and how that is dragging down the performance on the desktop. We have tried in vain to explain to you that the chipset is not the feature the drives the architectural platform to a NUMA memory model, the chipset is not responsible for scheduling execution on the CPU cores, and the chipset is not responsible for desginating the context of the thread and which node it should run through.

So at that I agree with the other posters, this thread should now be locked. It has bore no fruit.

Jack

You're not above me. You're just an anonymous voice on a public forum. I understand completely how the kernel and AMDs NUMA implementation work. That's why I'm confident that profiling can work along with "adaptive swapping." I'm sorry hat only regurgitation is available to you, but I am an engr. Teh type of person that makes the seemingly impossible possible.


And yes you did imply that no one would optimize for such a small-niche platform. I will not scan through the pages to find it.
If engrs felt like you there would be no Core 2. Nor any software optimization for any particular thing.

Give it up. It's a great idea and shows great promise for the multi-threaded future.

ah heck off chav.

To quote the British among us.

Baron, quit loading the forum full of BS again --- you are no engineer -- an engineer does not blatantly disregard data and draw erroneous conclusions like you do.

Also, I have never claimed specifically to be above you or anyone else for that matter, if you feel you are beneath me that is a subconscious observation injected into the conversation which would appear from your perspective.

Give it up.

Jack

Powerful you have become. The Dark Side I sense in you.
 
I think we should just ban Baron... that would solve everything.
Baron is like a dog that just causes too much pain... some time you have to just put them down, and I think now is the time 😀 :lol:
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom