New AMD CPU's

[jennidc]

Zuban - try and find an extra £40 and get the 8120 its a better chip - if you can wait a bit prices will drop a bit too

 

[Zuban]

i can wait as its a second pc but part of me just wants to sell the board and go sandybridge

 

[jennidc]

SB

I'm in the same boat although I can't help feel Sandybridge is a bit over hyped - good performance yes in Benchmarks etc but when I play Crysis and a mates Sandybridge plays Crysis even an expert would not know which system is which without going into control pannel - I'm a big Skyrim fan and it is supposedly CPU limited as the shadows are not done by the GFX card but by the CPU (thanks console ports!) but from my experience with the game both systems need to be set to Low/Medium shadows to run the game on Ultra without fps < 40.

As I am @ 4ghz on 6 core AMD for my next upgrade of CPU I need it to be 5ghz (or 6ghz Bulldozer) and have at least 6 or more cores for it to be a worthwhile upgrade so I will be running my rig for a while

 

[horrorwood]

I'm in the same boat although I can't help feel Sandybridge is a bit over hyped - good performance yes in Benchmarks etc but when I play Crysis and a mates Sandybridge plays Crysis even an expert would not know which system is which without going into control pannel

Yes really...

http://www.anandtech.com/show/4955/the-bulldozer-review-amd-fx8150-tested/8

How about...

http://www.techpowerup.com/reviews/AMD/HD_7970_CPU_Scaling/16.html

Bulldozer loses to the i7 920 in most games. The i7 920 was released in 2008.

AMD hyped up bulldozer to be amazing, yet you reckon sandybridge is over hyped? Lol

 

[DragonQ]

Bulldozer is far from a lost cause, even Zambezi has potential and already works well in certain scenarios.

AMD need to exploit their strengths whilst continuing tweaking Bulldozer. Plenty of time for them to improve their modular design by the time I'll need a new CPU.

 

[mmj_uk]

Potential doesn't mean much if they can't deliver on time, Intel development seems to be moving at a far faster pace than AMD... it's at the point now where even when Piledriver is finally released I don't have any confidence AMD will be any closer, they seem to be losing ground more than anything. It's so bad that we're still only comparing AMD's 8 core CPU's to Intel quads and completely ignoring their hex cores which could be released for for £250 at the drop of a hat.

 

[Gashman]

aye, the biggest problem they have is getting things from the drawing board to the foundries, that is their downfall. its not so much that their designs are bad, because they aren't at all bad its the fact that by the time the architecture sees the light of day, on silicon Intel have advanced a generation, based on original intentions Bulldozer should have been facing the previous generation of Intel processors, not Sandy Bridge and dare I say it is very competitive against the previous generation, dare I say even better. Bulldozer version II is was the version supposed (originally) to be locking horns with Sandy Bridge, but massive delays have put them almost an entire generation behind from plan, crippling their chances of knocking Intel off their throne.

920 or 8120, taking into account (hypothetical) they are both new processors, we know the performance, power consumption figures, temperatures, scaling and such of both, which would you consider to be superior? it isn't an easy decision and its much more difficult than comparing against Sandy Bridge.

also the intention since day one has been for one Bulldozer module to take on one rival core, the concept of adding two cores on ~10% more die space indicates the shift of focus from more per core performance to more cores. also cannot claim the decision doesn't have merit, against a 'stars' core the Bulldozer modules looses in single-threaded workloads but wins (quite well) in multi-threaded workloads, so the divide and conquer strategy can work well. don't compare Bulldozer in a core to core debate against Intel because that isn't fair, completely disregarding the fundamental architectural changes that happened in the design stage toward the 'more core' doctrine. its like saying 'one strong man, or two average men' which is better? it entirely depends on the work they are doing, in some situations the single strong man (Intel) is the best option for the task, but in others the two weaker men (AMD) are the better solution, neither is better than the other and both have their place in the world.

final note, 8120/8150 are supposed to be rivals to the 2500/2500K, not the 2600K with hyper-threading (which are priced quite a bit higher), comparing things from a market stand point. so comparing a Bulldozer against Sandy Bridge in price, it seems pretty competitive, cannot see why it would be regarded as otherwise?

 

[horrorwood]

final note, 8120/8150 are supposed to be rivals to the 2500/2500K, not the 2600K with hyper-threading (which are priced quite a bit higher), comparing things from a market stand point. so comparing a Bulldozer against Sandy Bridge in price, it seems pretty competitive, cannot see why it would be regarded as otherwise?

i5 2500k and the 8120 are basically the same price. How is the 8120 competitive?

 

[feeddagoat]

aye, the biggest problem they have is getting things from the drawing board to the foundries, that is their downfall. its not so much that their designs are bad, because they aren't at all bad its the fact that by the time the architecture sees the light of day, on silicon Intel have advanced a generation, based on original intentions Bulldozer should have been facing the previous generation of Intel processors, not Sandy Bridge and dare I say it is very competitive against the previous generation, dare I say even better. Bulldozer version II is was the version supposed (originally) to be locking horns with Sandy Bridge, but massive delays have put them almost an entire generation behind from plan, crippling their chances of knocking Intel off their throne.

920 or 8120, taking into account (hypothetical) they are both new processors, we know the performance, power consumption figures, temperatures, scaling and such of both, which would you consider to be superior? it isn't an easy decision and its much more difficult than comparing against Sandy Bridge.

also the intention since day one has been for one Bulldozer module to take on one rival core, the concept of adding two cores on ~10% more die space indicates the shift of focus from more per core performance to more cores. also cannot claim the decision doesn't have merit, against a 'stars' core the Bulldozer modules looses in single-threaded workloads but wins (quite well) in multi-threaded workloads, so the divide and conquer strategy can work well. don't compare Bulldozer in a core to core debate against Intel because that isn't fair, completely disregarding the fundamental architectural changes that happened in the design stage toward the 'more core' doctrine. its like saying 'one strong man, or two average men' which is better? it entirely depends on the work they are doing, in some situations the single strong man (Intel) is the best option for the task, but in others the two weaker men (AMD) are the better solution, neither is better than the other and both have their place in the world.

final note, 8120/8150 are supposed to be rivals to the 2500/2500K, not the 2600K with hyper-threading (which are priced quite a bit higher), comparing things from a market stand point. so comparing a Bulldozer against Sandy Bridge in price, it seems pretty competitive, cannot see why it would be regarded as otherwise?

2 bits of logic dont add up,
A) AMD and even a blind monkey could see intel wasnt gonna stand still and wait for this to come out. Intel have being carrying out a tick tock strategy since 06 almost to the dot with their release schedule. They would have known/figured out sandybridge's performance based on previous releases as well as moore's law.

B) bulldozer was slower than last gen in single threaded workloads, only marginally faster in multithread loads and from memory it did work well on floating point or did i swallow some marketing?

Amd needed a die shrink phenom to stay competive and sort 32nm out. It would allow bulldozer be released under a more mature process and possibley hit its targets. Amd learnt nothing from the switch to 40nm gpu node even tho they got that right.

 

[mmj_uk]

final note, 8120/8150 are supposed to be rivals to the 2500/2500K, not the 2600K with hyper-threading (which are priced quite a bit higher), comparing things from a market stand point. so comparing a Bulldozer against Sandy Bridge in price, it seems pretty competitive, cannot see why it would be regarded as otherwise?

Because BD stinks when you look at IPC which still continues to benefit the majority of applications released today, 8120/8150 would only be a match for 2500K in an ideal world where all applications would fully max out all available cores.

It's all well and good cherry picking a few encoding benchmarks where a fully maxed out Bulldozer holds its own against a fully maxed out 2500K/2600K and claiming it to be equal, the problem with that is it doesn't take into the account the majority of applications where BD's low IPC makes it a worse performer than even a Phenom 2.

 

[Martini1991]

Well, higher IPC inherently benefits everything really