AMD FX-Series "Bulldozer" will ship end of June

nehalem-ex is 8-core no? I bet in a single socket it'd be faster than BD 8core. Also, westemere-ex being 10 core would help. Perhaps in your company's situation a test vs nehalem-ex vs 8-core BD systems to really see. but i've heard that the AMD solutions tend to scale better in more sockets than one.
 
Your 99.9% number is wrong. And, in addition to that, what about the next few years? Aren't sw applications becoming more threaded every day? People made the same statement about dual core back in 2005. "No apps use more than 1 thread." Now today it is "no apps use more than 4 threads." What will it be in the near future? Seems that the "no apps use...." position is usually tied to intel's core counts more than anything else.

Eh, I've learned to buy computer hardware for the present, not the future. Buying for the future has always let me down. For 95%+ of enthusiast users in this current HW generation - I'm sorry, but IPC > #cores... And I can't wait to see what IPC increases BD brings to the table - I hope they're enough.
 
Last edited:
I bet in a single socket it'd be faster than BD 8core.

You sir, should be buying single core processors then. Unfortunately you are not my target audience. I can't be everything to everyone, it's the nature of the beast. I'd love your business, but you shouldn't buy something that doesn't fit your workload.
 
Dont get me wrong i want BD to be awesome just not sure on the desktop it will be any better than SB
 
I know there will be silicon improvements, but will Bulldozer have a higher heat tolerance before throttling as compared with the current generation? We'd still want to keep CPUs as cool as possible but knowing the construction is sturdier allows for some ease of mind when we see those upper 50s temperatures.
 
I know there will be silicon improvements, but will Bulldozer have a higher heat tolerance before throttling as compared with the current generation? We'd still want to keep CPUs as cool as possible but knowing the construction is sturdier allows for some ease of mind when we see those upper 50s temperatures.

It's impossible to know what kind of temperature loads Bulldozer can take until someone gets a hold of a chip or AMD releases documentation. Temperature limits depend on both the manufacturing process it's being built on and the architecture itself. But I wouldn't doubt it being able to handle higher temperature loads since it looks to be built with a high clockspeed in mind.
 
And I just upgraded to a Phenom II 955! I'm happy with it anyway, was an improvement over my old opteron 165. :D
 
This won't be backwards compatible with AM3 will it?

It will require a new motherboard that supports AM3+.

AM3+ will allow support for current AM3 chips but you can not put AM3+ chips in an AM3 motherboard.
 
Last edited:
It will require a new motherboard that supports AM3+.

AM3+ will allow support for current AM3 chips but you can not put AM3+ chips in an AM3 motherboard.

Ok, Thanks.

I might go to LGA 2011 then. The main appeal for me would be backwards compatibility, as my HTPC rig is AM3 based.
 
I assume you will be gaming on your HTPC. To me LGA2011 seems like an highly unusual and expensive choice for an HTPC. Although we will have to see. Hopefully bulldozer can compete with the higher end LGA1155 and at least match the lower end LGA2011.
 
I know there will be silicon improvements, but will Bulldozer have a higher heat tolerance before throttling as compared with the current generation? We'd still want to keep CPUs as cool as possible but knowing the construction is sturdier allows for some ease of mind when we see those upper 50s temperatures.

Power management will be better than the current generation, there are some new toys in there and more granular power gating.

Ok, Thanks.

I might go to LGA 2011 then. The main appeal for me would be backwards compatibility, as my HTPC rig is AM3 based.

That makes little sense to me. LGA2011 is going to be very large and very expensive. For an HTPC if you didn't want Bulldozer (and I could see that argument) then you might want to look at Llano instead. The integrated GPU will make it very power efficient and quieter for an HTPC.
 
Power management will be better than the current generation, there are some new toys in there and more granular power gating.



That makes little sense to me. LGA2011 is going to be very large and very expensive. For an HTPC if you didn't want Bulldozer (and I could see that argument) then you might want to look at Llano instead. The integrated GPU will make it very power efficient and quieter for an HTPC.

I was planning to turn my HTPC into a gaming rig by droping in a Bulldozer chip and a GTX 570, and then using my current aging gaming rig as an HTPC.But, as It turns out, I can't.

I overbuilt my HTPC, It's large enough to handle a GTX 570 and a Lga 2011 mobo. (Don't need it to be small, Theres a nice iron rack that can handle 100 lbs next to my TV)

In the end, I'm most likely just going to use my old gaming pc's case and PSU+ HDD+ Optical for the new gaming rig,sell the rest for pocket change either on FS/T or Fleabay, keep the HTPC untouched.
 
Xbit has some updated Bulldozer info:

http://www.xbitlabs.com/news/cpu/di...ulldozer_Chips_Incoming_Details_Revealed.html

There's 2 versions of the 8-core, one with TDP of 125w, the other 95w.

Memory subsystem is dual-channel 1866 MHz DDR3.

"AMD Internal Documents" project 8-core BD to have 50% improvement over PhII (I'm going to assume that AMD is talking about 6-core Thuban here) in multimedia applications.

I'm most interested in x264 encoding performance, personally speaking. If these projections are anywhere near reality, then based on my own napkin math, 8-core BD should be equal to 980x.

I just hope these don't cost a grand like the last FX-series of AMD chips.
 
They won't sell any chips at 1k. If the performance is that of the 980x, they'll price <400 and force intel to lose sales or lower prices.
 
^ 50% seems way too high, Even for well threaded applications.

If you take into consideration the 2 additional cores, that is 12.5% better performance per "core". Should the 50% boost be for when all cores are loaded and given that bulldozer cores will share some resources, single threaded performance should increase a bit more than that because shared resources will be used exclusively by one core.
 
If you take into consideration the 2 additional cores, that is 12.5% better performance per "core". Should the 50% boost be for when all cores are loaded and given that bulldozer cores will share some resources, single threaded performance should increase a bit more than that because shared resources will be used exclusively by one core.

I'm thinking 15% single threaded, 30% in wellthreaded apps.

I just don't see it competing effectively with LGA 2011.

If it was backwards compatible that would be A-OK, since I could keep my AM3 board. But since it'll cost me just as much as a new Intel rig, I won't use it unless it beats Intell in a fair fight, Something I haven't seen since the 478b days.
 
If you take into consideration the 2 additional cores, that is 12.5% better performance per "core". Should the 50% boost be for when all cores are loaded and given that bulldozer cores will share some resources, single threaded performance should increase a bit more than that because shared resources will be used exclusively by one core.

50/2=12.5?

Pardon me If I misread, but I just got mindfucked.
 
To get 50% more performance in 6 cores you need 9 cores assuming perfect scaling.

So 2 additional cores + 12.5% IPC improvement per core = the performance of 3 additional cores.

There was no talk about clock speed in this 50%.
 
Even if it closes the gap with SB there's still the question of whether it'll be able to reach the same lofty overclocks as the 2500k and 2600k.
 
To get 50% more performance in 6 cores you need 9 cores assuming perfect scaling.

So 2 additional cores + 12.5% IPC improvement per core = the performance of 3 additional cores.

There was no talk about clock speed in this 50%.

The Xbit article speculated speeds up to 3.5 GHz. I've seen similar numbers posted across the Interwebs. There's no mention whether this is the base speed or the Turbo speed, however.
 
What I meant by no talk about clock speed is what frequency is this 50% number against Phenom II compared at? Is the Bulldozer running at the same frequency as the Phenom II? What exactly is the benchmark? Too many variables from a leak like this..
 
So it's gonna be a war of AMD's "cores" vs Intel's "threads". Intel's Quad True Cores with Hyper-Threading are going to go up against AMD's Quad Modules that each have 3/4's of two cores. So when Intel releases a True Octo-Core, AMD is gonna strike back with a 16-core cpu. I think the core-wars are going to subside like the mhz wars did. It's gonna end up being Threads(which can refer to either the 80% Cores AMD has or the Real/Imaginary Cores Intel has) and power efficiency as the main marker of performance.

I'm really excited for Bulldozer and look forward to upgrading my 1090t to one in June. It'll probably run circles around my 2600k in encoding.
 
So it's gonna be a war of AMD's "cores" vs Intel's "threads". Intel's Quad True Cores with Hyper-Threading are going to go up against AMD's Quad Modules that each have 3/4's of two cores. So when Intel releases a True Octo-Core, AMD is gonna strike back with a 16-core cpu. I think the core-wars are going to subside like the mhz wars did. It's gonna end up being Threads(which can refer to either the 80% Cores AMD has or the Real/Imaginary Cores Intel has) and power efficiency as the main marker of performance.
:confused::(:confused::(
3/4's of two cores
Say what?
 
I want to see what a FX-8130P can do with 2 - 6990 in CFX.

(On maybe PCI Express 3.0 lanes if that's implemented by then.)

(with better cooling on the cards of course, + the 6990 drivers should be optimized by then as well)
 
Isn't the 2600k Intel's current top of the line for Sandy Bridge as well? There really isn't anything else they can position it against, at least until LGA2011 arrives.

No. 2600k is pure mainstream at $317.

Intel still has 2 price ranges above that, $583 and 999.If anything, AMD's positioning speaks volumes about what to expect performance wise.
 
No. 2600k is pure mainstream at $317.

Intel still has 2 price ranges above that, $583 and 999.If anything, AMD's positioning speaks volumes about what to expect performance wise.

Your right, provided this rumor is correct... Even though the high end socket 2011 i7's aren't out yet to compare them to, AMD could still compare them against the existing high end socket 1366 chips (e.g. 980x) that still outperform the current mainstream sandy bridge parts that were just released.

If the high end bulldozers performed anywhere near the high end Intel's, the bulldozers would be priced much higher. Maybe not as high as the equivalent Intel part, but a couple hundred more per chip wouldn't be unreasonable.

It could be a completely BS rumor. But if it's true, we could assume the high end BD parts perform near a sandy bridge 2600 series, or they perform much better and AMD wants to start another price war with Intel. The latter would be ill advised at this point, and not likely to happen since they would lose a price war.

If the high end BD parts perform on par with a 2600 sandy bridge and priced in the same price bracket, it would be both good and bad. They would be at the ideal price point for most enthusiasts, but they better overclock like fucking banshees for most enthusiasts to consider them over a 2600k that is already a proven monster overclocker.
 
Last edited:
I wouldn't call $317 pure mainstream. More like upper mainstream.

Not to mention i do believe that when you get right down to Oc'ing, if you max oc both the 980x and 2600k (within the normal achievable range) the 2600k does outpace the 980x on almost everything, am i not correct?
 
Back
Top