• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD..."Oh crap", Intel

Between Intel and AMD, Intel was first to 64-bit with IA64. But Intel was certainly not the first to 64-bit.
 
wonderfield said:
Between Intel and AMD, Intel was first to 64-bit with IA64. But Intel was certainly not the first to 64-bit.
Click to expand...

Wait, do Intel fans even want to admit to Itanium? That's kind of like a sin to speak about it...

But I get your point. Though the 64bit processes we use now in desktop computers are AMD's designs. I'm pretty sure Intel pays a license for its use.
 
The CPUs we use today are extended from the x86 ISA, which AMD licenses from Intel. Intel licenses the x86-64 extensions from AMD as part of a cross-licensing agreement.
 
Trine 2 @ 1280x720 same settings:

i5/4400 15 fps
AMD APU 72 fps

I'll pass on the slide show. Explains why new gen consoles went AMD.
 
It's true if you use integrated video, AMD blows away everything.
 
as many times as I heard AMD talk about some miracle fix, or a new CPU that is badass etc. Ill be honest don't think I could get excited about this.
 
4-core is boring and what I would build for my grandma so I'm glad AMD offers 8- and 16-core options for virtualization that are also price competitive.
 
When it comes to amd processors, keep expectations low and realistic. It's mainly on their gpu side where expectations can stay high.
 
jmilcher said:
This is news? Just more of the same. Hot air. AMD fan boys are so desperate to see AMD perhaps compete with Intel that they buy into this guy some how turning around Bulldozer.

The article also points out he is focusing on ARM but may "look at" x86. Not too much news there IMO.
Click to expand...

It is worse then hot air hot air has something this news article is full on conjecture and bias.

He will work on the K12 ARMv8-based products, but will take a look at Bulldozer as well and, maybe, finally allow it to get over the main problems.
Click to expand...

We all know what this says. Nothing absolutely nothing this is what you get when people start making up things.

What is clear is that Jim Keller is working on something they designed from scratch. To suggest that you bring an engineer to work on one important architecture then in his spare time fix Bulldozer because if it was worth fixing he would not be doing his job.

No one even knows if the new architecture will be cpu only part or APU?

Odds are higher that AMD will try to play it safe and promote the new Bulldozer cores as part of APUs. Steamroller gets Carrizo and Toronto (will come out this year), so Excavator will power the successors of those units.
Click to expand...

http://en.wikipedia.org/wiki/Excavator_(microarchitecture)

The Excavator based APU will be codenamed Carrizo, scheduled for a 2015 release.
Click to expand...

simple things go wrong when you don't even check what you are writing...
 
Last edited:
DASHlT said:
as many times as I heard AMD talk about some miracle fix, or a new CPU that is badass etc. Ill be honest don't think I could get excited about this.
Click to expand...

Nothing concrete has been issued. The only things to get excited about is that Jim is back.

He does command a lot of respect in the industry thou...
 
Well doing some reading and it appears they are going to be pairing x86 cores with ARM cores, so it's not ridiculous to think Keller is going to be tweaking the x86 cores.

I am going to cross my fingers.
 
Mackintire said:
Nothing concrete has been issued. The only things to get excited about is that Jim is back.

He does command a lot of respect in the industry thou...
Click to expand...

Just a mere chance they drop faildozer architecture and let it rot next to netburst cpus is more exciting than any other amd cpu news in last 3 years :)
 
It's going to take years for "his way" to be seen in the products. Anything released in the next couple of years are going to be based on previous to 2014 work.
 
If you read my other thread you'd know the details.

His jobs include assisting and aligning x86 resources for the skybridge architecture.

Add whatever small improvements he sees necessary to excavator, but NOT to impact the current schedule....which basically means he can do very little to excavator.

Design or redesign a new x86 core for 2016, not skybridge..... but elements of this project may be integrated into skybridge. This will be post excavator and may or may not include designs used on past architectures.

Given the time table I would not expect something 100% new.
 
At some point we should still be getting APUs based on ARM cores paired with Radeon iGPUs, correct? Given the TDP envelope available for desktops, could that combination provide something more than tablet level horsepower?
 
MrCaffeineX said:
At some point we should still be getting APUs based on ARM cores paired with Radeon iGPUs, correct? Given the TDP envelope available for desktops, could that combination provide something more than tablet level horsepower?
Click to expand...


Yes.

There will be ARM/GPU chips and ARM/x86/GPU chips. The will not run x86 based windows . Maybe windows RT?
 
well i'm a little excited by this and remembering my athlon 2400m system from way back when at the same time ha.
 
KazeoHin said:
Multi-core silicon? AMD did it first.
64 bit CPUs? AMD did it first.
Integrated memory controller? AMD did it first.
Click to expand...

Too bad that all happened all at once, not much has been done since then other then pushing more cores.

I did think it was funny with Intel's first Dual core being two physical cores, then next generation their quad core was two dual cores slapped together. :D

KazeoHin said:
Wait, do Intel fans even want to admit to Itanium? That's kind of like a sin to speak about it...
Click to expand...

Well there was a purpose for IA64 outside of general computing, it was small and useless, I'm surprised to see it last as long as it did.

Mackintire said:
Nothing concrete has been issued. The only things to get excited about is that Jim is back.

He does command a lot of respect in the industry thou...
Click to expand...

It's exciting enough to see him back, shows that AMD at least is trying.
 
I did think it was funny with Intel's first Dual core being two physical cores, then next generation their quad core was two dual cores slapped together.
Click to expand...

That was the straw that broke the camels back for me. I mean as a huge AMD fanboy constantly hearing all the crap about "true" quad cores and how great that was ... Meanwhile the "true" quadcores could not get the clocks up, had the TLB bug and underperformed.. Phenom I was a total disaster.
 
drescherjm said:
That was the straw that broke the camels back for me. I mean as a huge AMD fanboy constantly hearing all the crap about "true" quad cores and how great that was ... Meanwhile the "true" quadcores could not get the clocks up, had the TLB bug and underperformed.. Phenom I was a total disaster.
Click to expand...

Yea semantics aside, the Intel "quad" cores did perform very well.

Whats bad is I waited out to see the Phenom, stuck on a A64 somethingx2. I wanted to get a Q6600, but waited to see the Phenom. That pushed me to get a Q9450, which was nice that I waited, saved me some money but I would of still primarily been on AMD if the Phenom I wasn't so bad.
 
Did I jump through a temporal vortex or a teleporter or something, or am I the only one who realizes that Jim Keller rejoined AMD two years ago? This article reports on it like it literally just happened in May 2014 because that's when that conference featuring Jim took place. Coincidentally, development of the new K12 and unnamed x86 core began about two years ago, not long after Jim and some of his crew rejoined AMD, and said products will be introduced starting in 2016...

Much of this article makes little sense to me.

Come 2015, a new Bulldozer architecture called Excavator will be unveiled, Expreview reports. AMD will unveil a new general-purpose x86 core that should benefit from at least some of the great ideas that Jim Keller tends to get when he involves himself in a project.
Click to expand...

"New" in the sense that it hasn't been featured in a shipped product as of yet, sure, but Excavator the architecture has been known about for several years now.

For one thing, AVX 3.2 512-bit extensions, among other things, will be supported. Also, we may see up to 16 cores in a single processor.
Click to expand...

This is the first I've heard of the Excavator ISA supporting AVX 3.2 instructions. Last I heard, it would feature AVX2 instructions, along with a native 256-bit FPU capability which can do real 256-bit internal ops rather than breaking them down to 2x 128-bit ops internally. I'd take this report with plate-full 'o salt.

In fact, 16-core units could surface as early as this year (2014), once the Steamroller generation of Bulldozer comes forth. Sadly, software still isn't coded for so many different threads, so most of the potential computing power will go to waste, at least at first.
Click to expand...

As for "16 cores on a single processor", this is nonsense. That was a rumor from earlier this year which spawned off from an image from the AMD SWOG (software optimization guide), which depicted a hypothetical single-die processor with eight modules on it (16 cores) utilizing HT3.0 and being powered by the Steamroller uArch. The only way enthusiasts would see this is if they bought a G34 socket MOBO, as the package size cannot physically fit into C32 or AM3+. It is unlikely that this concept will ever actually make it into an product, since the roadmaps maintain that the server platform for G34 will be supported by Warsaw, which is still just two Piledriver dies on a single package.

Odds are higher that AMD will try to play it safe and promote the new Bulldozer cores as part of APUs. Steamroller gets Carrizo and Toronto (will come out this year), so Excavator will power the successors of those units.
Click to expand...

What odds are there when everyone already knows that the Excavator cores will literally only *be* in APU's and gimped fused-off-iGPU CPU derivatives, both on the consumer and server platforms? Furthermore, mind-blown at the terrible misinformation at the end there. Carrizo and Toronto are codenames for Excavator-powered consumer and server APU/CPU's, respectively, having nothing to do with Steamroller. The only successors to these chips will be the stuff coming in 2016, which will be totally unrelated to the Bulldozer family 15h processors.

Now we just have to wait and see if Jim Keller achieves the miracle that so many hope for and enables Bulldozer to match Intel Core i7 CPUs performance-wise.
Click to expand...

Even with the appreciable IPC increase that will come with EX, do not count on class-i7 performance per core/thread.
 
Last edited:
Yeah I was having trouble with this article as well.

I mean I would welcome any positive new things from AMD but this is so far out there that it looks something a conversation bot could come up with :)
 
daglesj said:
Hmmm what would I do?

Maybe build in a 1:1 Core to FPU setup.
Click to expand...

I would say instead of taking 125% of the transistors of a single phenom II core to make a 2 core bulldozer module use more like 150% of the transistors. 125% caused too many sacrifices for desktop loads that could not be easily made up by increasing the frequency.
 
Last edited:
daglesj said:
Hmmm what would I do?

Maybe build in a 1:1 Core to FPU setup.
Click to expand...

I don't think the shared FPUs have much impact. Some AM3+ boards have a 'one-core-per-module' mode, but it only improves single-threaded performance by 5-10%.
 
Kinda feels like the time when Microsoft invested 150 million in apple back in the 90's. They let all those guys go in order give Intel some competition.

Or... maybe this is a trojan horse.
 
DeathFromBelow said:
I don't think the shared FPUs have much impact. Some AM3+ boards have a 'one-core-per-module' mode, but it only improves single-threaded performance by 5-10%.
Click to expand...

Only if they could combine the integer execution units on each core of a module to form a strong single core with better IPC although that would be similar what Intel does with HT.
 
Last edited:
drescherjm said:
Only if they could combine the integer execution units on each core of a module to form a strong single core with better IPC although that would be similar what Intel does with HT.
Click to expand...

I take it you mean 'if only?' The 'one-core-per-module mode' I was talking about is a real feature on some AM3+ boards. It just disables half the cores on a given chip, eliminating performance degradation related to the shared FPUs and other components. It isn't used very often because you lose half your cores for a small performance boost. My point is that AMD's performance problem isn't due to the sharing of certain features on the modules, its that the integer cores themselves simply aren't very efficient.
 
I hope things really take off. Besides budget machines i have basically not touched AMD in years now.
 
DeathFromBelow said:
I take it you mean 'if only?' The 'one-core-per-module mode' I was talking about is a real feature on some AM3+ boards. It just disables half the cores on a given chip, eliminating performance degradation related to the shared FPUs and other components.
Click to expand...

No I was not talking about disabling cores / execution units but the idea of dynamically using the execution units of both cores inside a single module in a single thread when needed.
 
Last edited:
I don't see them beating Intel again. The Athlon caught Intel with their pants down and AMD rode it for years before Intel ditched P4 and went to the Core2 series and eventually i7/i5.

However, how ironic that AMD may have to ditch "faildozer" lol(borrowed from another poster). Long pipeline and low ipc, makes me feel like this is AMD's version of P4/Netburst lol.

Where Intel had their old P3/Centrino laptop chips that evolved into beasts, what does AMD have? I'm a HUGE AMD advocate, but adding Jim doesn't pull a rabbit out of the hat by default. I hope they can best Intel. Almost all my stock is in AMD and I've taken a huge beating over the past few years(bad moves in and out, but crap performance as well). I want AMD to put out good cpus and for the most part what they have now is ok, but just ok. I'm happy with my APU's for that I want them for, and my new 8320 is kind of a turd. Won't oc for crap(4.2ghz@1.45v) and I do see way more frame variance in gaming vs my i7 computer, but is amazing at some multi cpu workloads; so putting the right tasks to the right system is key. That said, if I had to buy just one cpu it would unfortunately be an i5/i7 at this time :(. Gotta give credit, where I wish I didn't have to, but Intel does own AMD right now.

My hope for AMD is that they can at least reasonably catch up to Intel in near performance, more give and take anyway like Nvidia/ATi graphics battles. They tend to trade blows in tasks until one trumps the other for a short time, then the cycle repeats. That would be acceptable I think to us all. When one, usually Nvidia takes the crown and keeps it for more than a year, then the prices and competition suffer. AMD/ATi have been able to catch up with Nvidia, so maybe there is a shot with the right talent they can close the gap with Intel.

Further, I'm not 100% sold that "faildozer" is totally doomed. It would take some engineering marvels yes, but I think that cpu has been crippled from the get go by crap lithography/processes and gremlins in the works. If they can iron out the faults and get the product out they wanted to, and then further evolve it, that cpu could be ok. But they've had to neuter it to get yields and now they have a Eunuch of a cpu. Also, if Microsoft got off their lard asses and made an OS that wasn't a craptastic turdevolution of their previous OS, bloated with junk and little on true innovation, then maybe the "faildozer" would show better as well. I'm thinking back to when AMD put out x86-64 and we waited for 2-3 years for an OS to show up that wasn't a joke. So I think in all, AMD doesn't have market power to lead MS, or others into creating products to make AMD cpus shine. AMD can't even make their products shine. Hopefully, they realize reality and create products that will show well with products that will be available in that product life cycle.
 
I just will say i really hope AMD puts real competition on the x86 scene. Of my last 5 computers, 2 were AMD and 3 were intel. Right now i wouldn't touch any AMD proc. But i also know we need actual competition to drive the market and i believe that so much i own amd stock even though i think their current products are mostly shit. I want them to do well.
 
Romale23 said:
I just will say i really hope AMD puts real competition on the x86 scene. Of my last 5 computers, 2 were AMD and 3 were intel. Right now i wouldn't touch any AMD proc. But i also know we need actual competition to drive the market and i believe that so much i own amd stock even though i think their current products are mostly shit. I want them to do well.
Click to expand...

There never was any competition. AMD is too small to matter. Even when the Athlon 64 was curb stomping Intel's products in every possible way Intel just kept on selling Netburst CPUs. It was cheaper to be anti-competitive than to rush out more competitive products.

Honestly, I don't think AMD's current lineup is all that bad. CPU speed generally isn't a bottleneck right now, the integrated GPUs on the FM2 chips are good, and I'd take a socket AM1 setup over an Atom. If you do a lot of multi-threaded work the 8-core chips are a very good value vs an i7.
 
I don't think AMD will compete again in x86. ARM is the future, and we even have a (shitty) version of windows running on ARM. Not to mention everyone and their dog are making ARM chips so it's not as much of a single-player-market as x86 is. Just because ARM is a RISC based arch does not mean it is slow, it's just that people are making slow ARM CPUs.
 
KazeoHin said:
I don't think AMD will compete again in x86. ARM is the future, and we even have a (shitty) version of windows running on ARM.
Click to expand...
My understanding is that ARM doesn't even come close to competing with the top-end of x86 single-threaded performance; certain markets care about this a lot.
 
KazeoHin said:
I don't think AMD will compete again in x86. ARM is the future, and we even have a (shitty) version of windows running on ARM. Not to mention everyone and their dog are making ARM chips so it's not as much of a single-player-market as x86 is. Just because ARM is a RISC based arch does not mean it is slow, it's just that people are making slow ARM CPUs.
Click to expand...

People have been saying ARM is the future for over 10 years now. They where wrong then they are wrong now.
 
You must log in or register to reply here.
Back
Top