AMD unleashes first ever commercial “5GHz” CPU, the FX-9590

[Zucker2k]

Comments like this are not helping honestly.

On 9590 (strange number somehow) I have no intention of buying one, but as others have stated at least AMD is trying. At least one company still has the balls to go toe to toe with Intel and stay in the game all these years.

How's this going "toe to toe"? At best, this is a desperate, half-ass*d effort aimed at the most clueless shoppers with money to burn. It could also be a "dry" run to pave the way for the 220w era on the desktop (perhaps the only way AMD can remain relevant on the desktop once Intel completes it's takeover of the igp battle; ie. 4 module, beefy igp with gddr5?). This will become evident once the benchmark results come in and the 4770k beats it handily with less than half the power consumption while costing a fraction! Also, marketing these as 5GHZ processors is like Intel marketing the 4770k as 3.9GHZ.

 

[Magic Carpet]

Preliminary reviews indicate the battery life improvements are impressive for laptops (even if this is an enthusiast forum) but the tradeoff is additional heat with no other real improvements. Perhaps Sony botched their design, but it can't be this bad.

http://www.laptopmag.com/reviews/laptops/sony-vaio-pro-13.aspx#review

That's just way too hot. Something's not right there. Sony is usually good at designs. Hmm.

 

[mrmt]

On 9590 (strange number somehow) I have no intention of buying one, but as others have stated at least AMD is trying. At least one company still has the balls to go toe to toe with Intel and stay in the game all these years.

Balls? I don't think it's a matter of balls, it's a matter of not have a choice. AMD has nothing else to do except try to compete with Intel.

 

[d3m]

Maybe it's 140W TDP.

http://mygaming.co.za/news/hardware/55134-amd-fx-9590-and-fx-9370-processors-unveiled.html

 

[Conscript]

Maybe it's 140W TDP.

http://mygaming.co.za/news/hardware/55134-amd-fx-9590-and-fx-9370-processors-unveiled.html

and right above that:

Below is a table that details what I expect the chips will match up to be.

It's a guess just like one anyone in this thread could make.

edit: not to mention, he absolutely has the base clocks wrong.

 

[inf64]

Yeah that author is clueless. He did write the disclaimer though...

 

[AtenRa]

The FX9370(4.4GHz base) could be 125W TDP, FX9590(4.7GHz base) could be 135-140W TDP.

 

[galego]

After car analogies, engineering decisions, and the occassional how awesome my Haswell is, let us return on topic.

The new centurion chips would be faster, in average, than i7-3770k. The 8350 at 4.6GHz gives a good idea of the waited performance for the FX-9590

http://openbenchmarking.org/result/1210227-RA-AMDFX835085&obr_shm=y

FX-8350: 35.95
i7-3770k: 33.14
FX-8350 @ 4.6GHz: 32.79

My guess is that the score of the FX-9590 would be about 30 points.

Regarding future games, the new centurion chips would perform at the level of Intel extreme chips

My guess is that the FX-9590 would achieve about 76 FPS.

 

[Chiropteran]

and right above that:

Below is a table that details what I expect the chips will match up to be.

It's a guess just like one anyone in this thread could make.

edit: not to mention, he absolutely has the base clocks wrong.

It makes sense what he says, though.

The extreme high power usage reported on FX-8150 was only when using a program that loaded all 8 cores, power usage while gaming or more regular activities didn't load all the cores, and was much more reasonable.

The author's theory is that the new turbo core technology that allows these CPU to clock up to 4.7 or 5ghz will only kick in for one or two cores. Power usage running 2 cores at 5ghz should actually be LESS than running all 8 cores at 4ghz, so this shouldn't require a massive bump for TDP.

 

[mrmt]

Help me out here, this comparison is largely nonsense, is that what you're saying? I'm honestly asking, because IDC is basically saying some of us are not smart enough (or engineers, or some such thing) to understand, so he tossed out a car analogy for us simpletons to grasp. Now I ask for specifics, because I am also an engineer, and I have a lot of hands on time with engines, so the comparison would be really easy for me if I have the numbers.

I don't know why you guys are so against car analogies, because from a business decision POV, the project trade offs in the MPU industry and in the car industry are exactly the same:

Should I spend X to get Y?

Yes, it is simple as that. This is the question the PM, the PM's boss and the staleholders ask every time someone comes suggesting something that will rise the project budget.

And this is what happened in Intel's case. After analyzing all alternatives with all engineering and costs inputs Intel went for the TIM we know. Of course they could have fielded a better solution, but it shouldn't be the best economic decision for them. So the answer to that question was no.

But instead of recognize this economic benefit for Intel and consider it a valid reason to go with the cheaper cooling solution, you are thinking that Intel is artificially impairing the performance of its products.

 

[GreenChile]

Also, marketing these as 5GHZ processors is like Intel marketing the 4770k as 3.9GHZ.

I didn't catch on to that one till you pointed that out. All other CPUs by Intel and AMD are marketed at their base speeds. This seems like a deceptive marketing scheme by AMD.

I hate it when companies do this sort of misdirection.

 

[GreenChile]

It makes sense what he says, though.

The extreme high power usage reported on FX-8150 was only when using a program that loaded all 8 cores, power usage while gaming or more regular activities didn't load all the cores, and was much more reasonable.

The author's theory is that the new turbo core technology that allows these CPU to clock up to 4.7 or 5ghz will only kick in for one or two cores. Power usage running 2 cores at 5ghz should actually be LESS than running all 8 cores at 4ghz, so this shouldn't require a massive bump for TDP.

But the base clock speed of all cores will still be at 4.7. This will require a massive bump in TDP unless AMD has some mystical voodoo they have found to solve the heat issue. This isn't likely because they would naturally want to waterfall it down to other product lines thus improving performance/watt and we haven't heard any such rumors.

 

[mrmt]

I didn't catch on to that one till you pointed that out. All other CPUs by Intel and AMD are marketed at their base speeds. This seems like a deceptive marketing scheme by AMD.

I hate it when companies do this sort of misdirection.

Did you expect less from AMD?

 

[SlowSpyder]

So the Ivybridge-E (socket 2011) 6 core i7's come out in 2 months. Of course with hyperthreading and all the same OC potential.

And my current ivybridge is already overclocked to 4.8ghz.... and far outperforms a fx8000 series of the same roughly overclocked equivalence.

So I'm failing to understand how this is good news... a fx8000 series chipped that's overclocked to slightly sub-5ghz range and eating about as much power as a highend GPU by itself, and struggles to compete with even i7's out of box that are quad core.

Is it the claim to fame for 5ghz? That's been done years ago... and pretty much done by any enthusiast who can walk out and pickup a cheap corsair h-series cooler today? Those "K" series processors? they are meant to run at whatever you can keep them stable and cool at. Any speed between their box rating to whatever suits you. You don't classify them as stock speed because they are sold as not being intended to run at stock speeds, or else you'd have just bought a locked model.

Seems like an attention grab that's going to come back to bite them. When intel showed off the 50+core "Knights Corner" 2 years ago, now that was impressive... that's something to watch. Lol, you want to talk about multithreading performance... there ya go. But this? I don't understand how this is any sort of great leap forward?

If you already have an Ivy at 4.8GHz, then this chip isn't aimed at you. You ask how this is good news, that AMD is releasing this, and say you don't understand how this is any sort of great leap forward. Its good news for AMD because the current FX chips aren't going to make them enough money to keep them in the game. These chips might not either, but sitting on the 8350 as their highest end is probably a shorter route to closing their doors. But I guess my question to you would be how is this bad news? If the CPU doesn't make sense for your needs, stick with your very capable Ivy, that's what most of us would do. I hope AMD prices these right, then maybe they can generate a little excitement over their CPU hardware again.

 

[guskline]

SlowSpyder, excellent post. Particularly the pricing. I saw someone post the pricing at $200 for the top end FX 9590. If that's the case AMD will be back in the game. If the converse is true ($800), I doubt there will be many takers. But hey, what do I know! $1,000.00 Titans appear to be flying off the shelves. I really wonder if this move by AMD is to buy time to work on a SteamRoller that really will change the landscape while trying to blunt the onslaught of Haswell? I wonder if AMD's high-end desktop cupboard is bare?

 

[frozentundra123456]

After car analogies, engineering decisions, and the occassional how awesome my Haswell is, let us return on topic.

The new centurion chips would be faster, in average, than i7-3770k. The 8350 at 4.6GHz gives a good idea of the waited performance for the FX-9590

http://openbenchmarking.org/result/1210227-RA-AMDFX835085&obr_shm=y

FX-8350: 35.95
i7-3770k: 33.14
FX-8350 @ 4.6GHz: 32.79

My guess is that the score of the FX-9590 would be about 30 points.

Regarding future games, the new centurion chips would perform at the level of Intel extreme chips

My guess is that the FX-9590 would achieve about 76 FPS.

Actually 76.23412.

 

[frozentundra123456]

It makes sense what he says, though.

The extreme high power usage reported on FX-8150 was only when using a program that loaded all 8 cores, power usage while gaming or more regular activities didn't load all the cores, and was much more reasonable.

The author's theory is that the new turbo core technology that allows these CPU to clock up to 4.7 or 5ghz will only kick in for one or two cores. Power usage running 2 cores at 5ghz should actually be LESS than running all 8 cores at 4ghz, so this shouldn't require a massive bump for TDP.

Yea, but if you are not using the 8 cores, quad intel will still have faster single or lightly threaded performance.

 

[Chiropteran]

Yea, but if you are not using the 8 cores, quad intel will still have faster single or lightly threaded performance.

Do you have benchmarks for an 8350 running at 5ghz to confirm this statement, or are you just guessing? I seem to recall that the 8350, at stock, was only slightly behind Intel, sometimes even ahead in a few rare cases. You really don't think a 20% increase in clock speed will make a significant difference?

But the base clock speed of all cores will still be at 4.7.

You aren't looking at the same image. The base clocks in the table referenced, which I was responding to, were listed as 3.7 or 4.0ghz, not 4.7.

Where did you hear the base clock would be 4.7ghz? That would just be silly. Why would a CPU with a max turbo of 4.7 also have a 4.7 base clock?

 

[inf64]

Cheaper one: 4.4/4.7Ghz
Pricier one: 4.7/5Ghz.

 

[Chiropteran]

Cheaper one: 4.4/4.7Ghz
Pricier one: 4.7/5Ghz.

Whats the source?

Anandtech: "We’ve asked AMD for details on the base clocks for the new parts, but so far have not yet received a response"

 

[guskline]

Chiroteran: In the "for whatever it's worth department" here are the links from AnandTech benchmarks for the 8350 vs the 3770K:
http://www.anandtech.com/bench/Product/697?vs=551

and the 8350 vs the 4770k:
http://www.anandtech.com/bench/Product/697?vs=836

Moving just the turbo speed for the 8350 from 4.2Ghz to 5Ghz while keeping the base at 4 Ghz will help some. However, if the rumor about increasing the base to 4.7Ghz and the turbo to 5 Ghz is true, then you would see significant improvement.

Since I have both my 3770k AND my 8350 OC'd I'll drop my 3770k back to stock this weekend and keep the 8350 at 4.6Ghz and run Aida64 to get some perspective as to what a base at 4.6 Ghz would mean for a 8350 vs a stock 3770k.

BTW, I seem to remember that when Bulldozer was first released, the knock on this architecture was that it needed to run at very high clockspeeds to be successful but that AMD had to "back off" the clock speed due to high heat and power issues. Perhaps AMD has found a way to reasonably control these matters now, thus 4.7Ghz/5ghz turbo.

 

[inf64]

Whats the source?

Anandtech: "We’ve asked AMD for details on the base clocks for the new parts, but so far have not yet received a response"

There is no official source. And the spec on the cheaper one is not yet set in stone (whether it's 4.4Ghz or something else). The other one is almost certainly going to have those specs I listed.

 

[blackened23]

That's arrogance at it's best. I have a degree but i don't think i'm better than the next person. As much as you're gifted you should turn down that "I'M BETTER THAN YOU DUMBA$$" attitude a couple notches.

On topic: in a way that's good for the desktop crowd since Intel seems to be slowing down speedwise

Well, IDC is human like the rest of us and it's just annoying to see cringe worthy posts constantly on here; I will say he has a higher tolerance threshold for BS than most, and makes informative posts all the time.

The salient point being - many of the comments are originating from those who are making broad assumptions about how things are done from an engineering standpoint which are entirely nonfactual. It's not like intel has engineering team meetings that go like this: "Hey gets lets figure out how to SCREW the overclockers! MWAHAHAHHA!" "VICTORY ACHIEVED!" (High-fives all around at Haswell's release). Come on. This is not how things happen from an engineering standpoint. There is no conspiracy at play here.

I've said it before and i'll say it again - chips that overclocked as well as SB did come once in a blue moon, and chips that are worthy of a yearly upgrade also come only once in a blue moon. I've been in the game for a while and I seldom recall CPU releases so groundbreaking that it made me *want* to upgrade faster than my typical 2-3 year upgrade cycle. There have been a few outliers, but overall Haswell fits in very nicely. It still overclocks well and has a very nice IPC boost over IVB and SB - a 4.3ghz Haswell is roughly equal to a 5ghz SB. Actually, i've seen benchmarks with a 4.3ghz 4770k outperforming a 5ghz 2600k. The Haswell is a great CPU, especially for the mobile sector - I really feel like people have a combination of inflated expectations and rose-tinted glasses from the SB. Again, once in a blue moon.

 

[GreenChile]

Where did you hear the base clock would be 4.7ghz? That would just be silly. Why would a CPU with a max turbo of 4.7 also have a 4.7 base clock?

I posted this source earlier. http://forums.anandtech.com/showpost.php?p=35130378&postcount=79

And 4.7 is supposed to be the base clock of the 5.0 max turbo version. The 4.7 turbo version has a base clock of 4.4.

 

[Torn Mind]

Well, AMD does have to extract every bit of cash it can, so why not? GHz is still marketable these days because it is not easy for the typical buyer to understand CPU speed and I would not be surprised if the Megahertz myth from the Pentium 4 days still carries on in a significant proportion of the population. From a business standpoint, releasing these chips make perfect sense. Get larger margins for a little bit more clockspeed, AND the other companies can stand to benefit very much as well.

Many people who are upgrading are prone to "overkill" or "future-proofing" and they WILL pour the money on these "boss-sounding" products. There's a reason ASUS sells Sabertooths; it's because people believe the "features" have substantial worth and shell out the cash.