Some FX-9370 results

Johan45

Limp Gawd
Joined
May 22, 2013
Messages
337
I've been playing around with the 9370 and I must say It is impressive for the fact I could hold 4.8 Prime 95 stable at 1.428v. The jump to 5.0 is big at 1.536v. I just noticed that I'm not able to attach images on here so here's a link to the SS of 5.0 and under neath are some benches. I wanted to see what it would do with just 1.5v and I was impressed.

http://www.overclockers.com/forums/showpost.php?p=7558898&postcount=589
 
then there is the power and heat lol, I still think AMD should have continued the phenom line and dabbled with the FX line as well, FX 8/6 for enthusiast, Phenom 6/4 for performance, APU for all around would have been a great lineup for them to go on, FX will be great at some point like APU is already starting to be however Phenom already was just needed further refining which would have helped them eek out what they could with the FX series as well.
 
If AMD would have done Phenom III and Athlon III, they would've already been close to Intel perf-wise eventually. Llano was based on the Stars arch, was die-shrunken down to 32nm, and was 6~7% faster than Deneb clock-for-clock.

However, I do enjoy the overall design of the modular concept, it's just the nature of these silicon microarchitectures that everything wouldn't be perfect since it was built from the ground up rather than being an evolution of a previous design. I firmly believe that, like dragonstongue said, FX will greatly improve its performance in time as the architectures have time to be refined and mature.

For comparison's sake, Intel is barely squeezing out more performance with each further uarch revision they do. It's basically a heavily-modified Pentium III-based design, after all. They can only do so much after such a long time. Since it seems like Intel plans to stick with that all the way until silicon itself tops out around 7 or 5nm, AMD will definitely catch up on the performance front. Jim Keller said they are on track to do it as well, and he is the last person I would doubt in this case.
 
Intel has always done this, they don't bother making jumps unless they have to, AMD is the innovater this is the problem many people with 1/2 a brain have been saying for a long time if Intel or Nvidia does not have AMD to compete with them they stop pushing the performance and shrinks as they will not need to do so then prices will go up soon you will be paying dearly for a very small bump with very little changes. Yeh AMD may not be top of the class in everything, but they keep the devils from making us all pay, this is why I stick with them, shame they didn't take the time to really make sure bulldozer was what it should have been and kept their mouth shut and just kept it behind closed doors so devs could have time to polish support for it.

I really think the design is cool very different concept however the way they market the FX 4-6-8 is not cool alas they will do what they can and if they are serious I can see another Athlon moment happening soon enough as AMD knows what Intel is doing, Intel can only see what AMD has done hence they stopped pushing the envelope(they are getting lazy) there is a very different approach that AMD and Intel take with their designs, AMD has the advantage on the type of silicon design but will suffer a power penalty though is a more robust design, Intel has power advantage and a more "simple" by design approach i.e bulk vs SiOn anyways, will be interesting to see what AMD comes up with to catch Intel offgaurd and certainly in the graphics side they have time to really focus on R-300 to face Maxwell.
 
I'm with you on that one dragonstongue, AMD may need some time to come around but I haven't lost faith in them just yet. Sure intel has the advantage right now but that hasn't always been the way and if AMD can pull it all together again, Intel will be the one catching up. We need them and I'm a firm supporter. I wanted to compare the Vischera line up and so far they're still great chips you just have to tame them. That is the part people aren't prepared for.
 
that and folks just want glory now, they want a works for everything type cure, a lot of folks also do not understand even the ones who have been using them for years that not a single chip out there is good for everything that and unless you compare a barely capable chip to a top of the line one the performance differences for *most* users are not like a "oh if you don't use this one you will just have bad performance and waste your monies" as so many people falsely convince others, it hurts us all in the long run, well, the ones who actually have to bust their ass to get somewhere with the little they have not so much for mommy and daddy kids who get it all or the ones blessed enough to be able to throw the $ away to "have the best" even if that best is super wasteful and makes the cycle continue.

I will continue on supporting the performance mid range which AMD seems to know best, even if top performance is not their agenda, value, performance, cutting edge always has been, not one maker out there knows how to do it all best, not one person, at least some try to not act as if they do, AMD IMHO is one of those even if their marketing team needs a good slap for spinning tales lol.
 
For comparison's sake, Intel is barely squeezing out more performance with each further uarch revision they do. It's basically a heavily-modified Pentium III-based design, after all. They can only do so much after such a long time. Since it seems like Intel plans to stick with that all the way until silicon itself tops out around 7 or 5nm, AMD will definitely catch up on the performance front. Jim Keller said they are on track to do it as well, and he is the last person I would doubt in this case.

Saying that a modern Core i series is "basically" a Pentium 3 from nearly 20 years ago is a great disservice to anything technology related. You may as well as make the argument that the modern internal combustion engine is barely any better then those made 100 years ago because they "basically" do the same thing. Don't forget electric cars were around 100 years ago too. Technology is all about making steps.

AMD got lucky 15 years ago because Intel got stubburn with Netburst. Once Intel pulled their head out of their ass and put their cash machine to work AMD had no chance of catching up.

With the release of the FX-9590, we are seeing almost the same thing that happened during the Netburst days, only the roles are reversed.
 
If AMD would have done Phenom III and Athlon III, they would've already been close to Intel perf-wise eventually. Llano was based on the Stars arch, was die-shrunken down to 32nm, and was 6~7% faster than Deneb clock-for-clock.

However, I do enjoy the overall design of the modular concept, it's just the nature of these silicon microarchitectures that everything wouldn't be perfect since it was built from the ground up rather than being an evolution of a previous design. I firmly believe that, like dragonstongue said, FX will greatly improve its performance in time as the architectures have time to be refined and mature.

For comparison's sake, Intel is barely squeezing out more performance with each further uarch revision they do. It's basically a heavily-modified Pentium III-based design, after all. They can only do so much after such a long time. Since it seems like Intel plans to stick with that all the way until silicon itself tops out around 7 or 5nm, AMD will definitely catch up on the performance front. Jim Keller said they are on track to do it as well, and he is the last person I would doubt in this case.

That would have cost significantly more, and AMD doesn't exactly have cash to throw around.

In addition, though IPC was higher in Stars, it overclocked poorly, with most not being able to hit 4 ghz on standard cooling if I recall correctly. Compared to that, Phenom II x6 were regularly hitting 4.2+ on standard cooling.
 
These cpus are great for video work or anything else that's multithreaded. They really shine. They're also a lot more tolerant to heat than their predecessors.
I have three of them and I work for a living dragonstongue. I'm at the point now in my life where I can spend more money and time on my hobby.
 
https://linustechtips.com/main/topi...rks-released-amd-takes-cpu-performance-crown/

AMD processors are really great for BF4. So when people start talking about how AMD needs to figure this out or work on this, I always wonder how much did the software developer care about making their engine 64 bit and multithreaded? DICE decided to multithread their game engine and this is the result.

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Battlefield_4_Beta-test-bf_4_proz_2.jpg


Not bad for a POS processor that can't do SuperPI right? Even the FX-8150 at stock speeds is holding it's own.
And look at the 1100T's score. The Phenom is getting old. A sad day indeed.

:)
 
it only took what 3 years? for a game design team to finally make use of my 8150 heh.
 
https://linustechtips.com/main/topi...rks-released-amd-takes-cpu-performance-crown/

AMD processors are really great for BF4. So when people start talking about how AMD needs to figure this out or work on this, I always wonder how much did the software developer care about making their engine 64 bit and multithreaded? DICE decided to multithread their game engine and this is the result.

[I MG]http://gamegpu.ru/images/remote/http--www.gamegpu.ru-images-stories-Test_GPU-Action-Battlefield_4_Beta-test-bf_4_proz_2.jpg[/IMG]

Not bad for a POS processor that can't do SuperPI right? Even the FX-8150 at stock speeds is holding it's own.
And look at the 1100T's score. The Phenom is getting old. A sad day indeed.

:)

Take those Intel X and K procs up to similar or at least closer clock speeds, not 40% different clocks, and watch the "holding its own" cease. When your CPU at stock speeds needs to have a 42% higher clockspeed to even hold its own, well I giggle.
 
westrock2000 said:
Saying that a modern Core i series is "basically" a Pentium 3 from nearly 20 years ago is a great disservice to anything technology related. You may as well as make the argument that the modern internal combustion engine is barely any better then those made 100 years ago because they "basically" do the same thing. Don't forget electric cars were around 100 years ago too. Technology is all about making steps.

Don't misconstrue what I was saying. I wasn't saying it was based off Pentium III as if it were a bad thing, I was saying that purely as a point of reference, and for comparison. I also said the modern variants are all 'heavily-modified', which is true. My reason for mentioning that was that AMD's newer arch was built from the ground up - hence - not everything was gonna be a smooth ride at the start. Intel didn't have to worry about that because all they have to worry about is further modifying and refining their pre-existing groundwork.

AMD got lucky 15 years ago because Intel got stubburn with Netburst. Once Intel pulled their head out of their ass and put their cash machine to work AMD had no chance of catching up.

With the release of the FX-9590, we are seeing almost the same thing that happened during the Netburst days, only the roles are reversed.

Don't be ridiculous. AMD were making competitive chips with Intel ever since the K6-I/II/III days, then with the Athlon, and Athlon XP, and A64. It's not like AMD were only up there with Intel for a nanosecond. It had nothing to do with AMD "getting lucky". I don't see why people seem to forget that AMD was never in a position where they had anything close to the amount of money and resources that Intel had, and were it not for IBM desiring two sources for their chips, AMD may not have even been around so long (at least in the x86 arena). Intel's prices would've been nightmarish.

As for the FX-9000, don't even waste time mentioning that. Those chips are nothing more than pre-overclocked Vishera's, there's nothing special about them. Piledriver has roughly the same IPC (I kinda hate this term, but whatever) as PII did, so of course it's behind Intel's modern stuff. Everyone knows this. It's been this way for quite some time.

Tsumi said:
In addition, though IPC was higher in Stars, it overclocked poorly, with most not being able to hit 4 ghz on standard cooling if I recall correctly. Compared to that, Phenom II x6 were regularly hitting 4.2+ on standard cooling.

This was more due to Llano's poor yields and GloFo's terrible transition to 32nm. They were having all sorts of manufacturing issues before these products released, which IIRC led to a small delay with Llano-based products initially. Bulldozer was also a less-than-stellar product which was based on GloFo's 32nm, but this also had the added bonus of being a fresh new uarch entirely.

Had they decided to further work on it, they could've made some impressive chips. However, I personally don't care about that because AMD wanted to go on with their CMT modular design and I like this newer design a lot. I just didn't bother with BD and PD due to some design concessions they made that I don't like, so I'm just waiting for Kaveri/SR, which is what BD should have been from the get-go.

Skripka said:
Take those Intel X and K procs up to similar or at least closer clock speeds, not 40% different clocks, and watch the "holding its own" cease. When your CPU at stock speeds needs to have a 42% higher clockspeed to even hold its own, well I giggle.

Who gives a shit? This is a stock clock comparison, and not only that, it's a POS unoptimized BETA. It was like this with BF3's Alpha and Beta's as well, the performance between those and the final game was a lot different. Regardless, a game like this shows how irrelevant the difference is between AMD and Intel is when it comes to programs that actually scale up to many threads. The 8350 is right there with the i7-2600k. You giggle at that? I laugh at you.
 
Geez I wasn't trying to start the INTEL VS AMD thing up. I just wanted to show some real world results.
 
I've been playing around with the 9370 and I must say It is impressive for the fact I could hold 4.8 Prime 95 stable at 1.428v. The jump to 5.0 is big at 1.536v.

Even if it could do 6GHz at 1.536v that's just too much heat to deal with. My 8320 could do 4.8 at 1.45v all summer without issues, beyond that I found I was running into thermal shutdown issues. It doesn't seem like you get anything for the extra cost of the 9370/9590.

Not that I dislike AMD. The FX8320 is a fantastic value at ~$160. I see very little real-world difference between it and my current Ivy Bridge-E setup. The 4930k's better IPC is noticeable in a few games, but not $420 more noticeable. I needed the second rig, but I kinda wish I'd saved some money and just bought another 8320.
 
Those benchmarks look shady to me. the 1100t has better IPC then the 8150/8350 but falls that far behind ? i would have to see it with my own eyes to believe it.

LL
 
1. Thuban doesn't really have higher IPC than Vishera. Vishera has pretty much the same performance, barring the module load penalty.

2. The 1100T has six cores, the 8350 has eight. Battlefield 4 is powered by Frostbite 3, which fully supports up to eight threads.

3. That graph is from the BF4 Alpha, which was far from optimized. Likewise this Beta version of BF4 is more optimized, but the final game will be more optimized than both.
 
Even if it could do 6GHz at 1.536v that's just too much heat to deal with. My 8320 could do 4.8 at 1.45v all summer without issues, beyond that I found I was running into thermal shutdown issues. It doesn't seem like you get anything for the extra cost of the 9370/9590.

Not that I dislike AMD. The FX8320 is a fantastic value at ~$160. I see very little real-world difference between it and my current Ivy Bridge-E setup. The 4930k's better IPC is noticeable in a few games, but not $420 more noticeable. I needed the second rig, but I kinda wish I'd saved some money and just bought another 8320.

I have a Fx-6350 in an ASUS M5A99FX Pro R2 that runs 4.6 GHz @ 1.416v on air cooling, very manageable.. This is a great combo I think for the price point and heat output. Very nice set up for an average user.
 
Take those Intel X and K procs up to similar or at least closer clock speeds, not 40% different clocks, and watch the "holding its own" cease. When your CPU at stock speeds needs to have a 42% higher clockspeed to even hold its own, well I giggle.

Yadda, yadda, yadda, overclock, yadda, yadda, yadda......:rolleyes:
 
Those benchmarks look shady to me. the 1100t has better IPC then the 8150/8350 but falls that far behind ? i would have to see it with my own eyes to believe it.

Only the 8150 had worse IPC. The 8350 has equal/better IPC, coupled with higher clock speeds and 2 additional cores.
 
and as we all know, at those res your more GPU bound than anything,,, get some SLI / XFire in there and watch Intel shine.

i was just considering a new AMD build but the power and heat just scare me!
 
Last edited:
Only the 8150 had worse IPC. The 8350 has equal/better IPC, coupled with higher clock speeds and 2 additional cores.

This...The X6's are still great CPUs, as long as you get them to ~4Ghz and more importantly, get the NB speed as close to 3Ghz as possible..The difference from 2Ghz to 3Ghz on the NB was insane in F@H, which is a great meteric since it scales VERY, VERY Well with threads/CPU speed/memory etc..

When I had my 1055T @ 4.3Ghz, it was a beast of a CPU...Once the X6's were clocked @ 4Ghz+, they were just as fast as the I7 Quads with HT (clocked at the same speed)...
I ONLY sold it because of an artificial limit imposed by Standford, the people that manage F@H..I would still be using it to this day in my 2nd machine otherwise...

I have wanted to play with another 8320/50, but AMD needs to refresh the 990FX platform if they end up keeping AM3+ alongside FM2+ (which I doubt will happen for much longer)...The AM3 socket has been around forever, and uses a good bit of power vs the FM2+/Intel setups..The PCI-E 16X 2.0 isn't that big a deal, since most Intel boards are only PCI-E 3.0 8X slots, which is the exact same bandwidth as the 16X 2.0 slots, unless you sell a kidney to go to the also semi-outdated X79 platform...

I have high hopes for SR parts, assuming they still have the O/C headroom the current chips do...A nice 30%-50% IPC improvement (according to AMD) over the current Vishera arch coupled with the same (at worst) or better power envelope should make for a nice alternative for the mainstream Intel platforms..

I think the recent high end FM2+ MoBo releases are a good sign of things to come..I just hope that AMD at least offers a hexcore option with an onboard GPU for HSA, and the OctaCore Variants without one (due to the massive die space it would need) so it doesn't end up being a part that is too costly to make a profit on, since most buying the OctaCore parts are going to be part of our crowd and use a dGPU...
 

Thanks for that, those prices would translate to about $350-400 AUD. (Not currency related but our shitty inflated tech prices). Guess I may as well buy a 3770k. I'm really getting sick and tired of Intel changing sockets and breaking incompatibility every few years when they feel like it.
 
Thanks for that, those prices would translate to about $350-400 AUD. (Not currency related but our shitty inflated tech prices). Guess I may as well buy a 3770k. I'm really getting sick and tired of Intel changing sockets and breaking incompatibility every few years when they feel like it.

If you are considering getting something new, then wait a few more months, since Kaverni is due to be released sometime this Fall/Winter..If the performance gains AMD is touting holds true, I would rather have one of them vs a 3770K (which I own myself)..It is going to be interesting to say the least to see what SR parts bring to the table.
 
Well I currently am using a GB Z68X UD5 + i3 2100 cpu I got off ebay for $60 few months ago and been wanting to grab a used i7 2600/2700k but the prices are too high here, damn flippers on flea bay. Was considering a 3770/3770k while the price was nice back in May when I built one of my clients a high end rig for WoW (3770/Asrock Fatality Z77 mono, 7950 OC) and the 3770 was a nice little performer.

But no, what happens is since the Haswell launch, Ivy Bridge i7 prices have gone up 30%. 3770k goes for $390 now rather than the $320 back then.

I really need something to replace the i3 2100 I have and I need some more grunt than a 3770k me thinks but the potential oc the chip can reach may compensate for the cores I need for work (came from a 48gb, 4.5ghz 980x setup). So it's either the AMD or a 3770k. Socket 2011 is out my budget.
 
Well I currently am using a GB Z68X UD5 + i3 2100 cpu I got off ebay for $60 few months ago and been wanting to grab a used i7 2600/2700k but the prices are too high here, damn flippers on flea bay....

Tell me about it. I have an EVGA Z68 SLI board sitting here that i bought a year ago (on a whim...i bought it before reading reviews that show it wasnt the best) and have been scouring eBay looking for some used SB's since the board's BIOS doesnt support Ivy...(and i dont want to bother with the hassle of buying an updated BIOS chip for Ivy support)

I found some local computer stores still selling 2500k's and 2550k's through online shipping only (not in there actual stores) but theyre still asking around 240 bucks.....about 10 bucks cheaper than a 4670k. Not sure if I should pull the trigger on a 2550k, or save up some cash and go Haswell, or migrate to a full AM3+/FX platform.
 
I have high hopes for SR parts, assuming they still have the O/C headroom the current chips do...A nice 30%-50% IPC improvement (according to AMD) over the current Vishera arch coupled with the same (at worst) or better power envelope should make for a nice alternative for the mainstream Intel platforms..

AMD never said anything above 30%, and the 30% gains mentioned at Hot Chips 2012 was over the initial Bulldozer, not Piledriver.

Here is an article which gives a really realistic look at what we could possibly expect from Kaveri: http://juanrga.com/en/AMD-kaveri-benchmark.html
 
I like kitguru's review of the 9590. He put it up against their 3960x OC'd to 4.4 and in the games they were literally neck and neck.

I may have missed it in the article that you linked, but why was the power consumption for the i7-3960X captured at 4.8GHz when the testing states that it was done at 4.4GHz? It makes the i7 look like it is drawing a lot more power than the FX under extreme load conditions, when in reality I would imagine that at 4.4GHz the i7s power draw shouldn't be that high.

I know from experience with overclocking my i7-920 that you can hit a wall with respect to power draw and heat that makes further OCing a game of maximum effort for minimum reward...

I never bothered to take full load measurements for the CPU only with my Phenom II 1090T overclocked rig, but now I wish that I had.

These newer FX chips are improving and my impression is that that can only be a good thing.
 
I may have missed it in the article that you linked, but why was the power consumption for the i7-3960X captured at 4.8GHz when the testing states that it was done at 4.4GHz? It makes the i7 look like it is drawing a lot more power than the FX under extreme load conditions, when in reality I would imagine that at 4.4GHz the i7s power draw shouldn't be that high.

I know from experience with overclocking my i7-920 that you can hit a wall with respect to power draw and heat that makes further OCing a game of maximum effort for minimum reward...

I never bothered to take full load measurements for the CPU only with my Phenom II 1090T overclocked rig, but now I wish that I had.

These newer FX chips are improving and my impression is that that can only be a good thing.

Not sure why they did that maube to give a clock per clock comparison. Either way I know these things really suck the power especially when clocked at the 5.0 and up range., I have them paired with 2 GTX 580's and a Corsair TX750w PSU wasn't enough when they were all cranked up.
 
Thanks for that yeah saw it last night. 9350 vs 3770k is the extra $70 premium worth over the AMD?
 
Thanks for that yeah saw it last night. 9350 vs 3770k is the extra $70 premium worth over the AMD?

Personally, if you already have a top-tier AMD mobo, the answer is obvious. If you are looking at a new system, though: The Intel will be about as fast/ faster but use HALF as much power and produce HALF as much heat.
 
when I can get one here in the US for 300 or under im getting one. I just tested bf4 beta with my 8350 and it kicks ass.
 
AMD never said anything above 30%, and the 30% gains mentioned at Hot Chips 2012 was over the initial Bulldozer, not Piledriver.

Here is an article which gives a really realistic look at what we could possibly expect from Kaveri: http://juanrga.com/en/AMD-kaveri-benchmark.html

That was a typo on my part, sorry I was tired..It was supposed to say 20-30%, over Vishera and Piledriver, respectively..Thanks for catching the error...

when I can get one here in the US for 300 or under im getting one. I just tested bf4 beta with my 8350 and it kicks ass.

Two things...First, since you already have a 8350, why not crank it up to 1.45V and see where it tops out at..Chances are it is going to be within ~100-150Mhz of the 9370 with ZERO cost to you...

Second, Your sig says that you have a Gigabyte 990FXUDA-5H, which is known to have some throttling issues with heavy loads on the 12V CPU rails (IIRC)...NewEgg states the following:

This CPU is only compatible with the following 990FX motherboards:

ASUS Crosshair V Formula-Z, GIGABYTE GA-990FXA-UD7 and ASRock 990FX Extreme9.
 
Last edited:
Back
Top