AMD FX Series Piledriver Processor IPC and Overclocking @ [H]

HvyMtl said:
But, is it fair to compare a $220ish processor to a $320ish processor?

True, this is an enthusiast website. However, not all of us have nearly unlimited cash to put in a system.
Click to expand...

Huh? My last two Intel CPU's - i5-2500k and i5-3570k were purchased for around $185 each. Even at stock speeds they slap down an overclocked AMD "$220ish" CPU with a 2x4 without even putting down their burger.

It's precisely the people without "nearly unlimited cash" that would want to be *avoiding* AMD from a cost per performance standpoint, unless ofcourse an emotional umbilical to AMD is holding them back.
 
Last edited:
odditory said:
Huh? My last two Intel CPU's - i5-2500k and i5-3570k were purchased for around $185 each. Even at stock speeds they slap down an overclocked AMD "$220ish" CPU with a 2x4 without even putting down their burger.

It's precisely the people without "nearly unlimited cash" that would want to be *avoiding* AMD from a cost per performance standpoint at this stage of the game unless an emotional umbilical to AMD is holding them back.
Click to expand...

Microcenter =/= everywhere else.

Also, the FX-8350 will crush the i5s and compete with the quad-core i7s in anything multithreaded, and there are plenty of productivity work that are multithreaded.

Once again, budget and usage dictates what you should buy, and AMD processors have their place.
 
This post is only for completeness purpose.

1. The US market, due to its unique scenario, volume,marketing arrangement and determination

2. Across most of the world markets, even many retail US sites, Intel Core i5-2500K is USD 220. Intel Core i5-3570K is USD 230. There are places where they are sold with slightly higher prices than listed here.

3. Across most of the world markets, even many retail US sites, Intel-based mid-range or premium motherboards are sold as such. There are places where they are sold with slightly higher prices than US retail price.

4. Certain immediate attention draw for some US users is due to the fact that, MicroCenters offer CPU/Motherboard bundle at a relatively steep discount that you can never get in other markets, due to special circumstances.
4.1 However, by design, you need to drive to MicroCentre. (remember two-way) If it is too far, you will incur higher gas-cost, accumulate car mileage and maintenance cost. Most people, statistically over large sampling populace, will stop along the way for food and drinks,maybe buy other things, so it benefits other parts of economy as well.
4.2 For some, you get time to travel outside. Some prefer moving around and fresh air. You may also join with family members, friends, hence potential social interaction. (depends on individuals)

5. Relatively, you can also get many good discounted price combination at many US online retail sites. Usually it is rather difficult to get the same level of discount at other non-US markets.

6. Again, nothing wrong, special marketing circumstances. However, in many other parts of the world, pricing difference is either as designated, or even higher than what you see here.

6.1 For example, across many parts of the world, you will never get USD185 Core i5-3570K.
6.2 When you calmly access the logical arguments, there are indeed very valid points and circumstances for Intel builds, even with the premium.
6.3 The same can be said of FX and Trinity APU builds, for example, if you must have AES/AVX/FMA due to software requirement. Hardware AES is not present on entry-level Intel processors currently.

7. Sometimes all the points are valid, but I only have 2 dollars..so you have to work with what you have...practically.
 
Last edited:
I am curious about win 8 performance, gaming and otherwise with the 8350 V Win 7.

Both x64 of course.

I will be upgrading to the 8350. How quickly and if/when I go to win 8 is all timey whimey.

If there is a noticeable performance boost with better thread handling it would be worth it for me.

I've heard Guild Wars 2 is very well threaded and uses as many threads as you have or at least more then two. Anyone in a position to check GW2 win 7 v WIn 8 on the 8350?
 
I miss the Jerry Sanders lead AMD. That guy was a fighter - and a bit of a gambler, in a good way.
 
Stop competing at the high end and becoming "low cost solution" it worked great for cyrix.
 
Hagrid said:
Depends on what they are competing at. :D
Click to expand...

Benchmark and e-peen competitions of course. What else is there? If you cant win benchmark competitions then your CPU is worthless and you wont even be able to boot into Windows. Its frustrating really cause Ive got all these great games and cant play any of them because of my damn CPU being so bad.
 
Piss poor methodology. You are expecting us to accept your rationalization for overclocking the Intel processors, while running the FX-8350 at stock frequency. Are you nuts??? YHpu compare then at stock frequencies or you overclock boith of them by the same amount of megahertz. Your method clearly give advantage to Intel.. No excuse for this at all. the fact is this FX has a greater uprange on overclocking than the IVY Bridge or the the Sandy Bridge cpus. For your games you could just have well chosen BF3, which is multithreaded, and couple of others that are better designed, as opposed to exclusively running poorly designed single threaded games. I don't see a virtue in rewarding poorly designed software and cpus that cater too poorly designed software, when there are viable alternatives that are well designed and execute better . When there are no viable alternatives that is when I can see the validity of your testing single threaded apps. They are the moribund past not the cutting edge of software development. So its as if you weight your tests ti insure the status quo remains the status quo. So your methodology for testing in my opinion is flawed , pretending objectivity but not coming close to it. The processor is not flawed but your testing mix certainly is. Single threaded apps should not be half or more of applicatrions tested and synthetic benchmarks should hardly be used at all. The dying refuse of software developers should be included in perhaps 20% of tests. There is winzip which is a far beter written program than winrar, It is is multtthreaded, yet you use winrar.whioch is single threaded. How much credit did you establish with Intel for this hack job?
 
Last edited:
No excuse for this at all. the fact is this FX has a greater uprange on overclocking than the IVY Bridge or the the Sandy Bridge cpus.
Click to expand...
Don't, like, all consumer chips have around the same practical overclock ceiling, around 4.5-5GHz depending on chip quality and cooling? So if the AMD chip starts out at 4.0GHz and can go to 4.5-5GHz that's a 500-1000MHz overclocking range or 12.5-25% overclock headroom, where a 3.4GHz intel chip going to 4.5-5GHz is a 1100-1600MHz overclock or 32.3-47% overclock headroom... For AMD to have the same "uprange on overclocking" they'd need to easily hit almost 5.3GHz on air and top out over 5.8GHz on decent water cooling setups with an above average chip. I don't recall any of the reviews where they hit even 5GHz easy and pushed up to 5.8GHz with some water and voltage tweaks.

I'm sure you'll counter with the fact that these chips, much like it's spiritual predacessor, the P4, overclocks extremely well at extreme tempuratures, to which my retort would be "who cares?" Nobody is going to sit there and pour LN2 into their case all day, doesn't seem you can even buy a vapochill unit any more, and judging by the fact that the last non LN2 pic posted in the Extreme Cooling Solutions forum here was from 2010 one might reason that sub zero solutions aren't exactly common or popular.

So considering the fact that the vast majority of overclocked AMD and intel systems will be operated within the same, again, practical operating range, comparing the two chips at the same frequency doesn't seem all that unfair to me.
 
There are plenty of reviews stating that they hit 5.0-5.2 ghz easily on high-end air (AIOs are considered air in terms of cooling capacity).
 
IB/SB should be compared to Vishera at stock clocks, not this overclocked Intel vs stock (perhaps even underclocked) AMD crap! I've been reading [H] for 15 years now and have supported a lot of the things that the people behind the site have done and have stood for, but things have changed since those days and the way these reviews are going down has finally disgusted me to the point where I can no longer support or recommend this as a "good" site for tech info (maybe for laughs or to piss people off). Stock for stock, Vishera actually performs better at the same price point than Intel in many benchmarks. That doesn't mean I recommend AMD over Intel, but it does mean that [H]'s comparison/testing methodology is absolute shyte! You lost a fan of 15 years. gg.
 
You know something I do have to admit that all the negative reviews are starting to get to me. I was watching this review at Overclock3D's Youtube channel and it really has me questioning my stance. Reading bad reviews all over the net is one thing but seeing this dude sit there for 38 minutes unscripted and see the pained, disgusted look on his face talking about Vishera is a lot more convincing than charts and graphs. Newegg has the 3570K and Asus Max V Gene combo for $411 shipped. Maybe its time I move on.

http://www.youtube.com/watch?v=u7wFKewAbgs&feature=share&list=UU_SN80_V2GymyCWM2oTYTeg
 
MacLeod said:
You know something I do have to admit that all the negative reviews are starting to get to me. I was watching this review at Overclock3D's Youtube channel and it really has me questioning my stance. Reading bad reviews all over the net is one thing but seeing this dude sit there for 38 minutes unscripted and see the pained, disgusted look on his face talking about Vishera is a lot more convincing than charts and graphs. Newegg has the 3570K and Asus Max V Gene combo for $411 shipped. Maybe its time I move on.

http://www.youtube.com/watch?v=u7wFKewAbgs&feature=share&list=UU_SN80_V2GymyCWM2oTYTeg
Click to expand...

Yeah, he makes good points, but it comes down to what you do and how much money you have to spend. I think you made a mistake upgrading from 1090T to Bulldozer. You 're spending too much money too quickly for marginal upgrades. Personally i am monitoring forums, hoping for someone with Thuban and a killawat, posts what his power draw is with Piledriver. Because i 've read so distant opinions about power draw, that makes me go insane...I mean, if this is true:

http://www.overclock.net/t/1319048/fx-8350-power-consumption-reduced-36

Then someone with a killawatt, please post, so that i can order it tomorrow!


EDIT: I think i will only upgrade if i happen to get my hands on a 8300 95W. I was thinking the FX-6300, but it would be a little slower than 1090T in video encoding, so does it really worth to give 140 euros for a bit less heat? Othewise i think waiting for Steamroller is more wise or keep piling up money to join the Empire and the dark side of the force eventually. Strong the lure is. AMD something must do! Only OEM the 8300 must be not!
 
Last edited:
os2wiz said:
Piss poor methodology. You are expecting us to accept your rationalization for overclocking the Intel processors, while running the FX-8350 at stock frequency. Are you nuts??? YHpu compare then at stock frequencies or you overclock boith of them by the same amount of megahertz. Your method clearly give advantage to Intel.. No excuse for this at all. the fact is this FX has a greater uprange on overclocking than the IVY Bridge or the the Sandy Bridge cpus. For your games you could just have well chosen BF3, which is multithreaded, and couple of others that are better designed, as opposed to exclusively running poorly designed single threaded games. I don't see a virtue in rewarding poorly designed software and cpus that cater too poorly designed software, when there are viable alternatives that are well designed and execute better . When there are no viable alternatives that is when I can see the validity of your testing single threaded apps. They are the moribund past not the cutting edge of software development. So its as if you weight your tests ti insure the status quo remains the status quo. So your methodology for testing in my opinion is flawed , pretending objectivity but not coming close to it. The processor is not flawed but your testing mix certainly is. Single threaded apps should not be half or more of applicatrions tested and synthetic benchmarks should hardly be used at all. The dying refuse of software developers should be included in perhaps 20% of tests. There is winzip which is a far beter written program than winrar, It is is multtthreaded, yet you use winrar.whioch is single threaded. How much credit did you establish with Intel for this hack job?
Click to expand...

They did overclock the FX, but instead of blindly following what AMD suggested, decided not to disable cores, which, IMO presents a more accurate representation of the CPU's overclocking capabilities i.e, [H] got it right. The point you need to see is that, it was not possible to OC both CPU's to have the same maximum clocks. As a consumer I want the following questions answered:
At my budget, what is the maximum stable performance one can extract by any means possible? Tie breaker: features offered by the platform.
Click to expand...

The question is, how do reviewers map sets of processors from both companies? One can ostensibly do it in the following ways:

  1. By price.
  2. By performance per Watt.
  3. By maximum performance boost obtained by overclocking.
  4. Arrange in sets of descending order and try a 1-1 mapping.
  5. By platform.
There are probably tons of other ways to build comparison tables and I'm sure you guys can do a better job than I have.


The point is, no single method listed above provides a complete characterization of competitiveness for the consumer. You really have to pick and choose what is important for you in the review (i.e.,the statement in the block quote will vary).


As for the rest of your statement regarding the tired old "software is poor" excuse, come on, be serious.
 
Super Nade said:
They did overclock the FX, but instead of blindly following what AMD suggested, decided not to disable cores,

As for the rest of your statement regarding the tired old "software is poor" excuse, come on, be serious.
Click to expand...

If somebody had reviewed the Pentium IV 2GHz using an AMD Athlon XP overclocked to 2GHz, there would have been a blood bath. Bad methodology. Either run both CPUs at default, or overlcock both to their respective maximums.
 
MacLeod said:
You know something I do have to admit that all the negative reviews are starting to get to me. I was watching this review at Overclock3D's Youtube channel and it really has me questioning my stance. Reading bad reviews all over the net is one thing but seeing this dude sit there for 38 minutes unscripted and see the pained, disgusted look on his face talking about Vishera is a lot more convincing than charts and graphs. Newegg has the 3570K and Asus Max V Gene combo for $411 shipped. Maybe its time I move on.

http://www.youtube.com/watch?v=u7wFKewAbgs&feature=share&list=UU_SN80_V2GymyCWM2oTYTeg
Click to expand...

I gave up 13 minutes into the video. :eek: In was constantly, 1090T this, 1090T that, 1090T the second coming crap. The chip that I sold is good, but, it was no better than the 945 I had in it before, I saw virtually no difference.

He keeps claiming miracle FSB speeds like 350 and 400, I say BS. The machine I have seemed to go a little slower on boot and other such stuff than the 945 I had in it before. I should have gone with a 8120 or 8150 right away but that was my mistake.

The 8120 I have now I see much better overall system performance. Things run smoother in my opinion and I have it up to 4.5GHz. I could not get a stable overclock above 3.8GHz on that 1090T no matter what I did. It cost me only $40 to go to the 8120 and I picked up the Thermaltake Water 2.0 Pro after that. (The H50 could not really handle an overclocked 1090T either.)
 
anubis44 said:
Either run both CPUs at default, or overlcock both to their respective maximums.
Click to expand...

If the only way to attain a maximum overclock on the FX is to disable a certain number of cores, then it is a bit unclear if the comparison is fair or not. The problem in comparing so many reviews is that there is no accepted definition of "apples to apples". Perhaps it would be a good idea to define this very clearly at the very outset. For example, one could state:

We are comparing processors that cost the same in two ways: a)without disabling parts of the CPU, b)Disabling parts of the CPU.
Click to expand...

Reviewers put a lot of effort in getting the data, but at times it is not clear what is it they are addressing?
 
ManofGod said:
I gave up 13 minutes into the video. :eek: In was constantly, 1090T this, 1090T that, 1090T the second coming crap. The chip that I sold is good, but, it was no better than the 945 I had in it before, I saw virtually no difference.

He keeps claiming miracle FSB speeds like 350 and 400, I say BS. The machine I have seemed to go a little slower on boot and other such stuff than the 945 I had in it before. I should have gone with a 8120 or 8150 right away but that was my mistake.

The 8120 I have now I see much better overall system performance. Things run smoother in my opinion and I have it up to 4.5GHz. I could not get a stable overclock above 3.8GHz on that 1090T no matter what I did. It cost me only $40 to go to the 8120 and I picked up the Thermaltake Water 2.0 Pro after that. (The H50 could not really handle an overclocked 1090T either.)
Click to expand...

Yeah maybe I'm just war weary about the whole thing. Tired of feeling like I have to justify why I've got a FX in my rig and reading that I've made a terrible mistake and my power bill is gonna go up $200 a month assuming I don't catch my wallpaper on fire and burn my house down and lets not forget none of my games will play either. I guess peer pressures is getting to me and I just want to be accepted. LOL :D

Going from a 1090 to this FX was an upgrade but only after I overclocked it 400 MHz faster than the X6. Definitely not worth it from a value standpoint but its been worth it from a standpoint of it being fun to overclock and tweak on and that's all I was really wanting.

The one thing I do call BS on with that guy is the whole "my D14 heatsink was hot to the touch with a 8150" line. I'm cooling one with a solid overclock with a $30 Hyper 212 and don't feel any blasts of heat radiating out of the case that sits 2' from me. I do think some of the doom and gloom is a little exaggerated with these.

At the end of the day I'm playing all my games at max (except BF3 which is GPU limited) and having fun overclocking so that's all that matters.....but it would be nice to be accepted by the in crowd just once. :p
 
Your power bill won't go up $200 a month regardless of the CPU you've got. My machines collectively can't do that.
 
Boy some people posting here have horrible reading comprehension. This is not a complete CPU review, it's an CPU IPC review. They are comparing Instruction Per Clock. WTF, people get a clue before you start commenting on their methodology.
 
Dan_D said:
Your power bill won't go up $200 a month regardless of the CPU you've got. My machines collectively can't do that.
Click to expand...

Just a little sarcasm brother. With Bulldozer and even Piledriver procs using 100 watts MORE than an Intel with a comparable clock, thats a lot of juice! ;)
 
Dan_D said:
Your power bill won't go up $200 a month regardless of the CPU you've got. My machines collectively can't do that.
Click to expand...

It's not just a matter of bill, but also a matter of heat. But anyway, i suppose you have cheap electricity in USA. Here depending on some factors it's 25 eurocents per kw on average. Then the more KW you consume, the higher some extra "power quotas" are and to everything you add VAT. I run 2 computers, one as HTPC let's say and one for video encoding, which is encoding 24/7. I can assure you that when i passed from a 45W to 125W, i saw the difference both in heat and electricity bill. I wouldn't mind bringing both down, as the more heat, will then require more cooling in summer.
 
Faethon said:
It's not just a matter of bill, but also a matter of heat. But anyway, i suppose you have cheap electricity in USA. Here depending on some factors it's 25 eurocents per kw on average. Then the more KW you consume, the higher some extra "power quotas" are and to everything you add VAT. I run 2 computers, one as HTPC let's say and one for video encoding, which is encoding 24/7. I can assure you that when i passed from a 45W to 125W, i saw the difference both in heat and electricity bill. I wouldn't mind bringing both down, as the more heat, will then require more cooling in summer.
Click to expand...

Electricity rates can vary quite a bit in different parts of the United States. The climate, the age of the buildings your in etc. all factor in. My house is new and built within all kinds of energy saving guidelines so it's pretty efficient. The house I was in is only about 7 years old and built when most of that stuff wasn't standard. While more efficient than the 60 year old house I was in even before that, it was larger so balanced out to being about the same. My power bill in the new house vs. the last one is about half what it was.

The single largest factor is typically heating and cooling with appliances being secondary. Computers and the home electronics rarely impact the bill all that much. I won't say there isn't a noticable impact, but the bill isn't altered nearly as much by turning off a computer or switching to one that consumes less power as updating your appliances with more efficient ones. In the winter time I rarely run the heat so whatever my summer bill is gets cut virtually in half again during the winter.

And yes I understand the arguments about heat. I sit in an office with 5 monitors, 1 laptop, two relatively high end machines plus whatever is on the test bench running all the time. It's easily 10 degrees warmer in this room then it is anywhere else in the air conditioned living space of the house.
 
I'm toying with a 6300. I also own a 960T that doesn't unlock.

In Skyrim using the straight ultra preset in cities like Markarth the 960T @ 4.2GHz will match the 6300 @ 4.8 in min FPS. They also share the same numbers @ 4.0GHz and 4.6GHz respectively. Running around Markarth loads one core @ about 96% and another at roughly 60%, so IPC is important.

The 960T @ 4.2GHz and the 6300 @ 4.8GHz share the same score in cinebench only using one core. It's worth noting the 960T I have will allow me to run my northbridge @ over 3000 and the 6300 I have can't hold 2600.

I haven't bothered to compare them in many multithreaded apps because the 960T doesn't unlock so the 6300 will be faster. The 6300 @ 4.8GHz gets around 8226 KB/s in WinRAR Beta 2's bench.

So if you replace an 1100T|1090T @ 4.0GHz with an FX 8320 or 50 you will likely need on the order of 4.5GHz to match it in lightly threaded apps. In heavily threaded apps you should easily pass it.
 
Last edited:
Faethon said:
It's not just a matter of bill, but also a matter of heat. But anyway, i suppose you have cheap electricity in USA. Here depending on some factors it's 25 eurocents per kw on average. Then the more KW you consume, the higher some extra "power quotas" are and to everything you add VAT. I run 2 computers, one as HTPC let's say and one for video encoding, which is encoding 24/7. I can assure you that when i passed from a 45W to 125W, i saw the difference both in heat and electricity bill. I wouldn't mind bringing both down, as the more heat, will then require more cooling in summer.
Click to expand...

The heat isnt that bad. That is the one thing that is way overstated in reviews. Im sitting literally 2 feet from my case and dont feel any extra heat coming from my overclocked 8150. Hell, Intel procs run into the upper 70's when overclocked while AMD procs are in the upper 50's and I dont see any Intel owners crying about their rooms being overheated or their heat sinks hot to the touch. Some of the stories about FX procs are exaggerated.
 
MacLeod said:
The heat isnt that bad. That is the one thing that is way overstated in reviews. Im sitting literally 2 feet from my case and dont feel any extra heat coming from my overclocked 8150. Hell, Intel procs run into the upper 70's when overclocked while AMD procs are in the upper 50's and I dont see any Intel owners crying about their rooms being overheated or their heat sinks hot to the touch. Some of the stories about FX procs are exaggerated.
Click to expand...

Well, to tell the truth, i haven't seen how a Bulldozer behaves in summer, but i 've read various horror stories about how hot they get, specially in oc (the guy in the video also speaks of nuclear reactor!), so Bulldozer scared me enough just for that. I really don't know why Intels get so hot in the core, but if i am not mistaken TDP is about how much heat the cooler must dissipate. So Intels are rated much less, but they get hotter in the core. Maybe they 're desiged to keep more heat inside and so less heat comes up to the surface? I don't know. But all reviews show Intels drawing much less power from the wall. And physics say that the more you draw, the higher the heat that will dissipate per Joule effect.
 
Dan_D said:
Electricity rates can vary quite a bit in different parts of the United States. The climate, the age of the buildings your in etc. all factor in. My house is new and built within all kinds of energy saving guidelines so it's pretty efficient. The house I was in is only about 7 years old and built when most of that stuff wasn't standard. While more efficient than the 60 year old house I was in even before that, it was larger so balanced out to being about the same. My power bill in the new house vs. the last one is about half what it was.

The single largest factor is typically heating and cooling with appliances being secondary. Computers and the home electronics rarely impact the bill all that much. I won't say there isn't a noticable impact, but the bill isn't altered nearly as much by turning off a computer or switching to one that consumes less power as updating your appliances with more efficient ones. In the winter time I rarely run the heat so whatever my summer bill is gets cut virtually in half again during the winter.

And yes I understand the arguments about heat. I sit in an office with 5 monitors, 1 laptop, two relatively high end machines plus whatever is on the test bench running all the time. It's easily 10 degrees warmer in this room then it is anywhere else in the air conditioned living space of the house.
Click to expand...

Yes, unfortunately, i live in late 19th century building in "historical center", where even the cables aren't as efficient as today's and the building has as you imagine virtually no built-in power saving technology for either winter or summer. But the worst is summer, as we hit 32-35C with high humidity (sea is very close). The building is such build that you can't even put air conditioners in every room nor can you cover all rooms, as ceilings are unusually high by today's standards and this alone reduces efficiency. The computer room itself is poorly ventilated, so at summer you can feel heat very easily and the 2 computers don't help.
 
Great review, as usual, [H]! Thanks for taking the time.

AMD, here's hoping that you have the R&D in progress to close that performance gap and really lower the power draw.
 
Faethon said:
Well, to tell the truth, i haven't seen how a Bulldozer behaves in summer, but i 've read various horror stories about how hot they get, specially in oc (the guy in the video also speaks of nuclear reactor!), so Bulldozer scared me enough just for that. I really don't know why Intels get so hot in the core, but if i am not mistaken TDP is about how much heat the cooler must dissipate. So Intels are rated much less, but they get hotter in the core. Maybe they 're desiged to keep more heat inside and so less heat comes up to the surface? I don't know. But all reviews show Intels drawing much less power from the wall. And physics say that the more you draw, the higher the heat that will dissipate per Joule effect.
Click to expand...

Yeah but like I said, Ive got one right here at eye level barely 2' away and running IBT or Prime95 it does put out heat but not the amounts youre reading about in all the reviews. I maintain that its very much exaggerated. Maybe its just because Ive got a Hyper 212 and not something higher and and its not capable of radiating a large amount of heat away from the proc.
 
MacLeod said:
Yeah but like I said, Ive got one right here at eye level barely 2' away and running IBT or Prime95 it does put out heat but not the amounts youre reading about in all the reviews. I maintain that its very much exaggerated. Maybe its just because Ive got a Hyper 212 and not something higher and and its not capable of radiating a large amount of heat away from the proc.
Click to expand...

Honestly, i think Steamroller is too far away and i have an itch for upgrade. But on the other hand, i am not sure if it's logical to upgrade to Piledriver. Ideally heat-wise, i 'd prefer an 8300 95W, if i could nail one (hard to know). But would it really be a good upgrade? I suspect i would get at best 2fps increase in encoding ( i fear less). Which is a bit "meh". From upgrade point of view, it would make more sense to go for 8320 or 8350, but i am more afraid of heat in summer...

So i am torn. Another side in me says "maybe it's better stop throwing good money after bad and take the decisive step and bow to the Empire", but i 'd hate to waste my investment on AM3+.

I don't know. I will probably wait to see what more users will say once they test Piledriver. I am also scared about eventual problems with programs or Windows, cause with the Thuban i must say that i have zero problems, i can't make Windows BSOD once even if i wanted.
 
XamediX said:
Boy some people posting here have horrible reading comprehension. This is not a complete CPU review, it's an CPU IPC review. They are comparing Instruction Per Clock. WTF, people get a clue before you start commenting on their methodology.
Click to expand...

Yes, I see that. My concern is that this is the only review of the FX-8350 on here. What is the point of an IPC review anyhow? On a gaming/enthusiast site like [H], shouldn't they be more interested in the maximum gaming performance of the chip and how it compares to other chips in and around the same price range? No, there's no law against this kind of review, but it smells suspiciously like an attempt to portray the new AMD chips in the worst possible light. I know from Kyle's rants since the release of Bulldozer that he's not happy with AMD, but for crying out loud, does every review of an AMD product have to be an anti-AMD rant? Yes, I too lament the refocusing of AMD's efforts towards a future, multi-threaded software universe a litltle bit earlier than would be ideal, but that doesn't mean AMD is wrong. All the games tested in this 'review' were not only notoriously single-threaded but, they are based on very old x87 code for floating point calculations. These games are designed for a single-core, pre-SSE/AVX/FMA world that simply doesn't exist anymore.

AMD has already explained they've seen the shift happening in CPUs to becoming essentially integer processing monsters and floating point being moved almost entirely to GPUs, so that's what they've spent their limited R&D designing: an 8 core, integer processing titan, and that's what the Bulldozer/Piledriver architecture is:

Since I can't paste the link because it keeps deleting the domain name portion, you'll have to go to benchmark reviews.com and click on the "PassMark PerformanceTest 7.0" page. The FX chips are so good at multi-core aware, integer performance that the guy writing the article said: "The FX-8350's performance on the Integer section of the benchmark is amazing...so much so that I must chalk it up to a problem or compatibility issue between the FX-8350 and this particular benchmark." What we'll find is that this isn't a mistake. The AMD architecture really IS that fast at integer calculations when the code is multi-core optimized. Steamroller or Excavator will finally integrate Radeon cores into the same die and complete the process. At that point, virtually all FPU will be done by either the integrated FPU Radeon cores, or on the add-in GPU board.

Just say "NO" to crappy, single-threaded, x87 software. If we all do it, we'll finally get properly optimized applications and games.
 
Last edited:
X87 benchmarks should NOT be the defining point of a review they are not relevant
 
Tsumi said:
There are plenty of reviews stating that they hit 5.0-5.2 ghz easily on high-end air (AIOs are considered air in terms of cooling capacity).
Click to expand...

OK some AMD chips have a small amount more headroom than I gave them credit for, frankly I didn't scour the internet looking at reviews, just hit a few of the big sites and saw 4.9-5.0 (assuming [H]'s 4.6 was a bad chip). I was responding to the guy who was claiming AMD had more overclocking headroom than intel chips, which even at 5.2GHz is still false, because it's looking like 5.2-5.3GHz is about the max for non extreme cooling methods, and those numbers aren't anywhere near the headroom that intel chips have. Intel is WAY conservative on their binning compared to AMD as is readily apparent in their power usage.

Even if we get super generous and say average AMD OC will end up around 5.0-5.4GHz that still gives us:

AMD FX-8350 @ 4.0GHz stock: 1000-1400MHz OC (25-35%)
Pick your intel poison @ 3.4GHz stock: 1100-1600MHz OC (32.3-47%)

So being super generous using bigger numbers than I could find from any review (that didn't use LN2) AMD's overclocking headroom high end still barely overlaps intel's low end.
 
-Dragon- said:
OK some AMD chips have a small amount more headroom than I gave them credit for, frankly I didn't scour the internet looking at reviews, just hit a few of the big sites and saw 4.9-5.0 (assuming [H]'s 4.6 was a bad chip). I was responding to the guy who was claiming AMD had more overclocking headroom than intel chips, which even at 5.2GHz is still false, because it's looking like 5.2-5.3GHz is about the max for non extreme cooling methods, and those numbers aren't anywhere near the headroom that intel chips have. Intel is WAY conservative on their binning compared to AMD as is readily apparent in their power usage.

Even if we get super generous and say average AMD OC will end up around 5.0-5.4GHz that still gives us:

AMD FX-8350 @ 4.0GHz stock: 1000-1400MHz OC (25-35%)
Pick your intel poison @ 3.4GHz stock: 1100-1600MHz OC (32.3-47%)

So being super generous using bigger numbers than I could find from any review (that didn't use LN2) AMD's overclocking headroom high end still barely overlaps intel's low end.
Click to expand...

And when IPC is factored in the headroom of the Piledriver / Vishera chips means even less. So even if they could beat the Intel's by 200-400MHz that's not enough headroom to beat Intel's performance in the applications which the FX-8350 do better in. And the gap between Intel and AMD is only going to widen as Haswell and Ivy Bridge-E show up.
 
You must log in or register to reply here.
Back
Top