Negatives about amd cpu

[TheToucan]

Currently I'm weighing the pros/cons of getting a fx 8320 with a r9 390. Assuming ill mainly be web browsing/gaming what are the negatives about picking the 8320 over an i5 4690k.

 

(Side note- ill most likely be upgrading to skylake or whatever Intel CPU is the newest at the end of the year).

[Samfisher]

Currently I'm weighing the pros/cons of getting a fx 8320 with a r9 390. Assuming ill mainly be web browsing/gaming what are the negatives about picking the 8320 over an i5 4690k.

 

(Side note- ill most likely be upgrading to skylake or whatever Intel CPU is the newest at the end of the year).

Slow.  Almost everything a regular user will use a PC for depends on single threaded performance, games included, which happens to be AMD CPUs weakness now.

[FruitBasketSilex]

Pcie3/M.2/Bottlenecking/Heat/Single-Core_Performance/FPS_in_games

[tahmidkhan123]

they are of a really old architecture which renders them inferior to intel processors in almost all applications. Only go AMD if your budget does not allow for anything else. 

[bmacsterGAMING]

Yeah don't go AMD. If you are willing to shell out a few extra bucks, definitely go Intel. 

[Dabombinable]

Well for starters its recommended that you don't go over an R9 280 with an FX8320, though the 280X should be fine at a stretch. AMD's CPU have been severely lacking single threaded performance since at the very least early 2006 while only just matching in multi threaded performance CPU with the same base core/module and thread count-and only at a significantly higher clock speed, they are all highly inefficient, they are all outdated by a few years technologically speaking, and outside of a few tasks they are poor value for money.

[theereal]

Cant really think of a benefit for going with AMD Instead of Intel In this case

[WoodenMarker]

Cant really think of a benefit for going with AMD Instead of Intel In this case

Lower price compared to the 4690k. That doesn't mean much though since Intel still has similarly priced alternatives that are just as good. 

[Sky Daddy]

as someone who went from an oc'd fx8320 (4.7ghz) to an i5 4690k, i can't really recommend it for gaming unless you don't mind  you're games randomly stuttering

 

sure it will run games fine, but you will get much more noticeable fps dips in cpu bottleneck scenarios 

 

ie cpu related fps dip on stock i5 4690k ~60fps on fx 9590 ~36 fps (napkin math i5 having 40% single threaded performance)

 

this will vary from game to game but in all cases the i5 is going to perform the same or better

[warzkaz]

lack of mini itx motherboards

[RefurBuilds]

AMD CPU's even being older are still very safe bet to go with gaming.If you are on a budget than definitely better go for AMD system than Intel.

 

However if you are planning an upgrade, get a decent Z97 motherboard which again will be costly and get that i5. In the future you will have an upgrade path to i7

Generally the difference you see in specific games, because more GPU bound games will perform similarly between those too (there might be the thing that AMD will get MIN FPS lower), but for more CPU bound games of course Intel is better.

 

But yet again I would go cheap with AMD if I would be playing on Full HD 60Hz monitor...  The Intel system is to pricey for that kind of generation  

[Dabombinable]

as someone who went from an oc'd fx8320 (4.7ghz) to an i5 4690k, i can't really recommend it for gaming unless you don't mind  you're games randomly stuttering

 

sure it will run games fine, but you will get much more noticeable fps dips in cpu bottleneck scenarios 

 

ie cpu related fps dip on stock i5 4690k ~60fps on fx 9590 ~36 fps (napkin math i5 having 40% single threaded performance)

 

this will vary from game to game but in all cases the i5 is going to perform the same or better

I think its well over double at the same clock speed if my Phenom II N970 is anything to go by when compared to an i5 540M-and the Phenom II (K10) has a higher IPC than the FX8320 (Vishera)

[Tytan64]

I really enjoy watching his videos and he made one about that topic. I hope you enjoy it as well =)
 

[Prysin]

Well for starters its recommended that you don't go over an R9 280 with an FX8320, though the 280X should be fine at a stretch. AMD's CPU have been severely lacking single threaded performance since at the very least early 2006 while only just matching in multi threaded performance CPU with the same base core/module and thread count-and only at a significantly higher clock speed, they are all highly inefficient, they are all outdated by a few years technologically speaking, and outside of a few tasks they are poor value for money.

i still cant find the place where it AMD officially say that R9 280 is the reccomended max for FX8.

[Tytan64]

I'm trying to get it documented but I remember AMD itself using Intel processors to show off what their new GPUs were capable of doing two years ago because their processors were simply not powerfull enough. So at a certain level you can expect them to bottleneck high end GPUs. In my opinion this is just the case for GPUs over 600$.

I wouldn't pair a Titan X with an AMD processor but a R9 390X is still fine in my opinion.

[Dabombinable]

i still cant find the place where it AMD officially say that R9 280 is the reccomended max for FX8.

Its what they advertised when they released the e series, ergo its what they recommend. And the recommendations do fit with user experiences.

[Prysin]

the e series are massively underclocked. they only turbo if they have super low temps.

a 8320e loses to a 8320 non-e in every fucking benchmark

http://anandtech.com/bench/product/1402?vs=698

in gaming benchmarks its even worse. because not all games are able to trigger turbo on AMD CPUs. The load they put on the CPUs simply arent heavy enough for turbo to engage all the time. Unlike intel which turbos almost all the time.

here. 8370e vs 8320

http://anandtech.com/bench/product/1340?vs=698

8370e has higher turbo speeds then 8320. so if it turbos, it should win, every fucking time.

the 8320e that AT got was a golden sample. it hit 4.8Ghz AND was possible to undervolt from stock.... that is fucking insane. So even a golden chip vs the 8320 and stock vs stock IT STILL LOSES.

yet when we do 8320e @ 4.8Ghz things are different. VERY different.

http://anandtech.com/bench/product/1403?vs=698

their 8320e at 4.8Ghz beats out a 9590. Thats how good their chip was/is.

in gaming a 8320e is equal or WORSE then a FX 6300.... yeah, thats how bad it is.

those CPUs are meant as an upgrade for older/lesser chipset users, like 880 and 970 chipsets. Yet in practice, their mostly equal due to clock speed being notably lower then normal.

IF however, you spend the time to OC these e chips, they will offer higher clockspeeds at lower voltages. AND THAT IS ONLY IF YOU BOTHER TO MANUALLY OC AND UNDERVOLT.

[Prysin]

Its what they advertised when they released the e series, ergo its what they recommend. And the recommendations do fit with user experiences.

also what user experience?

your own?

so far, people who bitch about stuff is usually the type who can be described as "grass is always greener on the other side".

been there, done that. guess how much it cost me. 550 USD.

guess how much performance i gained. 20% in games.

guess how much my old FX8320 bottlenecked the 295x2 OUTSIDE of The Witcher 3 (which didnt have CF profiles, and also, the reason i changed to intel was to play TW3 and heavily modded skyrim at full details)

the answer is> not a whole lot...

ill ask my friend if i can have my FX and sabertooth back for some testing over the coming weekend. he still havent built his rig.

then ill redo some benches with my FX running at its proper speeds.

ill have to mod my kraken x60 to life again, or run stock cooler with the FX though...

Intel, so far, has provided me with 3 benefits with their z97 chipset.

PCIe Gen3

M.2

Support for high speed memory.

what did i lose vs my old 990FX chipset?

4 USB 2 ports

2 USB 3 ports

2x eSata ports

tripple CF

2x Sata ports

So i lost shitloads oc connectivity. Which is neat, since i own 2x external HDDs, webcam, mouse, KB + some more.

was it worth spending 550USD to get a i7? no. it wasnt.

my old setup ran 2x HD 7950 in CF without issues. Yes i tested them with my i7, they perform a few FPS better, but that is more likely to be due to recent driver improvements as on average, i only got like 5FPS more in games i KNOW can use all 8 threads of a CPU.

so stop spreading bullshit. you havent owned a FX, you never will.

just like with FM2+ vs i3. If you did a blind test without any FPS counters, just pure gaming with no evidence of which was which. i dare say you wouldnt know which was which.

Remember, just because a game uses ONE core, doesnt mean the load is heavy. The CPU workload itself is usually pretty light for games. Although intense, it is not complex. Complexity is what kills FX. Synthetics uses complex code. Thats why intel will win. the workload itself exposes FXs weakness. Which is good, because a synthetic benchmark should do that.

however, let me ask you this.

Haswell is, core for core, 70% faster then FX Piledriver cores.

so in a game that is purely CPU bound with a TitanX, running at 600x480 we should in theory see 70% higher performance.

BUT WE DONT. we see at most 30%.

so what is all that superior IPC doing? What is the lower cache and sub-cache latencies doing?

Or more importantly, WHY THE FUCK IS FX PERFORMING THAT GOOD???? it should be smashed, rightfully so. It loses, but nowhere near as bad as it SHOULD.

now, i will not try to claim AMD has exchanged 70% core performance so they could add some vodoo or witchcraft into their architecture. But in the end, when you boil it down. Most of the benefit Intel has over AMD is PURELY IN YOUR HEAD.

once you get over the "giddy" period where "everything new is so much better", you realize, its just like it was.

so at this point, i am more interested in finding out why Intel processors perform so badly compared to FX.

Because any way you look at it. a 970 is a 970.... if you have 70% more raw CPU power, that should by all means produce MORE then 10fps difference....

to give you an idea how much power that is..

skylake i5 6600k vs Sandy Bridge i5 2500k -> skylake is 30% stronger, core vs core. And it yields roughly 30% better performance when both are using RAM with equal speed. So why the fuck can FX which is notably slower, by every possible measurable way, be within 30% of ANY FUCKING INTEL CPU???

[KuJoe]

The downside to picking an AMD FX-8320 over an Intel i5-4690K is deciding what you want to spend that extra $90 on.  :lol:

[Nena Trinity]

Price vs single threaded performance?

[ChrisCross]

upgrading in a year doesnt really makes sense. i would go with skylake now because if you go with an old architecture now you will need to buy new ddr4 ram for the new system and you end up wasting a lot of money. just safe a bit and go for skylake now.

 

if you dont want to do that here are pros and cons:

-amd cpus and mobos are cheaper but have less single threaded performance

-meaning despite having less computational multithreaded performance the i5 performs better in games

-intel has more pci lanes so you can put more gpus and high bandwith ssd's in there (btw skylake is way better than haswell another reason to get it now)

[ApolloX75]

snips

You raise some excellent points and I've got some extras observations to add.

So before I switched up to Haswell I was running a QX9650 on a 790i board. Make no mistake in modern games it was bottlenecking my 970 by about 30-40%. So when I switched to my 4670k the difference was night and day in modern games.

However there is an interesting point to this. While I was running the QX for fun I benched a lot of things. Including 3D Mark 06 and other "antiquated" benchmarks since they released them all for free and I was feeling nostalgic. After I upgraded to the 4670k none of those benchmarks would run properly, and some wouldn't run at all like the CPU rendering tests.

I've tried repeatedly to get them to run out of curiosity in different environments and still nothing. I don't know if this is universal or whatever, I haven't taken the time to increase my sample pool. But it strikes me as odd that newer "superior" hardware has issues with simple synthetic benchmarks from the past, and only in the CPU test department.

Just some food for thought. I have a hankering to run some tests on the FX 6300 system in the house, and I'll bet those old benchmarks will work fine on it.

[Prysin]

You raise some excellent points and I've got some extras observations to add.

So before I switched up to Haswell I was running a QX9650 on a 790i board. Make no mistake in modern games it was bottlenecking my 970 by about 30-40%. So when I switched to my 4670k the difference was night and day in modern games.

However there is an interesting point to this. While I was running the QX for fun I benched a lot of things. Including 3D Mark 06 and other "antiquated" benchmarks since they released them all for free and I was feeling nostalgic. After I upgraded to the 4670k none of those benchmarks would run properly, and some wouldn't run at all like the CPU rendering tests.

I've tried repeatedly to get them to run out of curiosity in different environments and still nothing. I don't know if this is universal or whatever, I haven't taken the time to increase my sample pool. But it strikes me as odd that newer "superior" hardware has issues with simple synthetic benchmarks from the past, and only in the CPU test department.

Just some food for thought. I have a hankering to run some tests on the FX 6300 system in the house, and I'll bet those old benchmarks will work fine on it.

to make Haswell run old 3D mark tests you must enable "3D mark legacy mode" in the BIOS

[ApolloX75]

to make Haswell run old 3D mark tests you must enable "3D mark legacy mode" in the BIOS

It's definitely not just Haswell. Even FX CPUs struggle with older software tests. Just confirmed with the 6300, it won't crash but can only average 1 FPS.

[a3venom]

The biggest negative is single threaded performance, which even 8 cores can't make up for.

An i3 6100 beats a FX overclocked 8350 in gaming in most of the titles (ofcourse the FX wins some) and they are the same price.

 

With the i3 you can buy a $50 Mobo and no need of a CPU cooler , but with the FX you need atleast a nice mobo if not any cooler, so technically, the AMD chip is worse value for money.

 

It has been that way for some time, AMD has ceased to be a viable alternative even for budget builds. It is only good if you want to do encoding and heavy multitasking, otherwise it is pure garbage to buy right now.

 

Plus it offers no upgrade path, whereas you have a decent upgrade path the intel way.

 

People who say that intel costs more money are just misinformed now