vishera single thread performance VS

[justin4pack]

Ok so this is a simple question, i hope to get a simple answer. From all my reading and understanding, single thread performance of something like the fx-6300 is way under performance of say the i5-2500k. BUT in multi thread they are the same if not 6300 a bit better. So if i overclock to say 4.2GHZ on the 6300 will the single thread performance match that of a 2500k? I know this seems a moot point but if someone could just confirm this that would be great.

 

[SlowSpyder]

You can overclock a 6300 to reach 2500k performance. The thing is that you can also overclock the Intel chip, which reaches levels of single threaded performance Vishera cannot touch. But if your goal is to get to ~stock clock 2500k performance, you should be able to get there roughly.

*edit - the 2500k turbos to 3.7GHz. If you figure it has a ~30% IPC advantage (I think?), you'll want to clock high enough to make that up.

 

[ShintaiDK]

When single thread, it actually includes all cores. So in compare to the i5 2500K it would be up to 4 threads. But beyond 3 threads, the FX 6300 gets the module penalty. So for the FX you get something like 100% performance for 2 threads and 80% for 2 more if 4 is used.

And the FX-6300 is not better in multithreaded either:
http://anandtech.com/bench/Product/699?vs=288

But the 2500K is pretty old by now

 

[sm625]

There are many single threaded tasks where no plausible amount of overclock can make up the difference. The most painful thing would probably be javascript. Most audio encoding is going to be really bad since most are still single threaded. If you're using iTunes, then just forget it.

 

[ruhtraeel]

When single thread, it actually includes all cores. So in compare to the i5 2500K it would be up to 4 threads. But beyond 3 threads, the FX 6300 gets the module penalty. So for the FX you get something like 100% performance for 2 threads and 80% for 2 more if 4 is used.

And the FX-6300 is not better in multithreaded either:
http://anandtech.com/bench/Product/699?vs=288

But the 2500K is pretty old by now

Sandy Bridge = OC'able
Ivy Bridge = + 5-15% performance, Intel being cheap and using paste instead of solder = higher temps = less OC'able
Haswell = + 5-15% performance, Intel still being cheap and using paste but also has the voltage regulator on the chip = even higher temps = even less OC'able

Result = Little to no improvement over 2 generations, which makes me sad, considering the jump from Nehalem to SB.


I'd imagine OC'ing the FX series would do it well to make up for single thread performance, while keeping its multi thread performance decent.

 

[know of fence]

[del]

 

[SlowSpyder]

Also, contrary to what was once popular opinion, it seems that overclocking the NB/L3 on the FX chips can certainly add performance.

 

[galego]

Ok so this is a simple question, i hope to get a simple answer. From all my reading and understanding, single thread performance of something like the fx-6300 is way under performance of say the i5-2500k. BUT in multi thread they are the same if not 6300 a bit better. So if i overclock to say 4.2GHZ on the 6300 will the single thread performance match that of a 2500k? I know this seems a moot point but if someone could just confirm this that would be great.

Yes, when the six cores are in use the FX is faster, but the i5 has about a 30% better performance per core.

You would OC the FX to something like 4.5GHz to get the same single-threaded performance, but 4.2GHz is enough, because you would obtain about a 93% of the single thread performance of the i5, whereas increasing the gap in multithreaded.

Check also the next link for extra advantages of the FX chip

http://www.cpu-world.com/Compare/434/AMD_FX-Series_FX-6300_vs_Intel_Core_i5_i5-2550K.html

 

[SPBHM]

http://www.cpu-world.com/Compare/434/AMD_FX-Series_FX-6300_vs_Intel_Core_i5_i5-2550K.html

this is misleading, how can you really consider lower clock a disadvantage when it gives you higher performance and lower power usage?!

as for ST performance, you will need a lot of overclock, like for most things 5GHz is not going to be enough http://www.hardwarecanucks.com/foru...166-amd-fx-9590-review-piledriver-5ghz-6.html , and once you start doing high overclocks with one, and leave the other at stock settings... the stock will have an advantage in cost and reliability... so the stock i5 is probably way better value.

also as said before the i5 have a BIG advantage in up to 4 threads performance (so most things people do with their PCs), not only "single thread".

 

[BigChickenJim]

I actually use an FX-6300 at 4.0 GHz in my current rig. The short answer is no, it will not rival an i5 2500k or better in gaming performance as of yet (this could change soon given the new consoles' reliance on AMD multi-core architecture). The longer answer is that real-world gaming performance differences between an OC 6300 and an i5 are not nearly as massive as many benchmarks would have you believe. In-the-wild, end-user gaming experience differences are much harder to detect when you don't have 50 bar graphs showing you every minute variation.

Intel makes better processors without question, but I still don't always think the performance gain is high enough to justify the extra cost if you're operating on a budget.

 

[bgt]

The only thing that matters is...how does it perform in real life? Do you notice the single threaded performance advantage Intel has? I doubt it. If I sit behind my 2 main desktops, a 3770K and a 8350 I only see the FX being snappier than my i7. Especially after powering your pc, coming into desktop and instantly start the explorer. With the i7 you have to wait 15-20 seconds before you can surfe anywhere. Just a small example.
The whole experience with the 8350 is just awsome....it never waits,stutters. With my i7 thats a different story. Too many apps open and it will start waiting longer.

 

[galego]

The only thing that matters is...how does it perform in real life? Do you notice the single threaded performance advantage Intel has? I doubt it. If I sit behind my 2 main desktops, a 3770K and a 8350 I only see the FX being snappier than my i7. Especially after powering your pc, coming into desktop and instantly start the explorer. With the i7 you have to wait 15-20 seconds before you can surfe anywhere. Just a small example.
The whole experience with the 8350 is just awsome....it never waits,stutters. With my i7 thats a different story. Too many apps open and it will start waiting longer.

You are here reporting the famous lag problem of the intel i7.

 

[guskline]

bgt, does the I-7 "lag" happen if you use the 7950 in the 3770k?

I have both the 3770k AND the 8350 with the same ssd boot drive and do not notice lag in either one. galego, what is the "famous" I-7 lag all about?

 

[bgt]

bgt, does the I-7 "lag" happen if you use the 7950 in the 3770k?I have both the 3770k AND the 8350 with the same ssd boot drive and do not notice lag in either one. galego, what is the "famous" I-7 lag all about?

Guskline, the lag is at its highest when just booting up and immediately opening explorer, it has nothing to do with the graphic card. After that its smooth. But again sometimes opening too many apps while busy unrarring iso's f.i it stutters a bit. Thats why I changed from a 2500 to a 8350. The 2500 was even worse to a degree that it was becoming a drag. This happened most when my Windows was already a long time on my pc and a lot of installed programs were there. If I then had 4 iso's busy unrarring and downloading a big file f.i the CPU could even stall.

 

[Ed1]

The only thing that matters is...how does it perform in real life? Do you notice the single threaded performance advantage Intel has? I doubt it. If I sit behind my 2 main desktops, a 3770K and a 8350 I only see the FX being snappier than my i7. Especially after powering your pc, coming into desktop and instantly start the explorer. With the i7 you have to wait 15-20 seconds before you can surfe anywhere. Just a small example.
The whole experience with the 8350 is just awsome....it never waits,stutters. With my i7 thats a different story. Too many apps open and it will start waiting longer.

I don't have AMD system to compare .
I would guess what your seeing is on Intel CPU/MB combo Intel loads a few processes once you install all drivers which I guess AMD doesn't have .Many of these you could probably disable on home system if it bothered you , UNS, LMS and there a few services .
I think its more how win7+ starts up to desktop fast but is still loading process in background , some with delay built in .

 

[Durp]

You are here reporting the famous lag problem of the intel i7.

What? Is there any substance to this or is this just the usual Galego amdzone type of thing?

 

[sm625]

This "lag" is obviously an issue with intel chipset drivers. Soemtimes it takes a while for a driver to load. If the post-load performance is better then its a tradeoff most are willing to accept.

 

[ShintaiDK]

What? Is there any substance to this or is this just the usual Galego amdzone type of thing?

Last part.

 

[Ed1]

This "lag" is obviously an issue with intel chipset drivers. Soemtimes it takes a while for a driver to load. If the post-load performance is better then its a tradeoff most are willing to accept.

Yes, even if there is a slight wait for all DLL to load I much rather have faster system when I run apps . My guess on this forum many don't even reboot/restart daily .

 

[parvadomus]

The Intel CPUs have like 30%-40% better IPC in most single threaded workloads. However AMD CPUs run at a higher frecuency, as they are engineered to run that way.

 

[galego]

What? Is there any substance to this or is this just the usual Galego amdzone type of thing?

Last part.

In this same thread one user is reporting how his i7 performs poor than his FX. But please ignore the lag problem and go with the usual personal attacks.

 

[CHADBOGA]

In this same thread one user is reporting how his i7 performs poor than his FX. But please ignore the lag problem and go with the usual personal attacks.

Yes it is amazing how all the members of AMDZone, when they try an Intel system, always experience problems of a bizarre nature that simply aren't encountered by anyone else.

Truly amazing.

 

[galego]

Yes it is amazing how all the members of AMDZone, when they try an Intel system, always experience problems of a bizarre nature that simply aren't encountered by anyone else.

Truly amazing.

Before you lacking any data that supports your unwelcome attack on him (specially when supplemented with your signature), your post is not going to eliminate the lag problem.

 

[SPBHM]

Before you lacking any data that supports your unwelcome attack on him (specially when supplemented with your signature), your post is not going to eliminate the lag problem.

it must be a problem with his specific PC, so I guess he can test other parts, RMA his bad stuff or something,

because if you really believe a core i7 would cause performance issues during the boot process and not something else (software? SSD/Hardrive?) and the 8350 wouldn't

that's why you have to test parts seriously and not go by this sort of thing.

and I remember the exact same user, before even buying the i7 talking in some topic about how he didn't wanted to buy the i7 because he didn't like Hyper Threading...

it looks to me he made his mind about his favorite parts way before giving the i7 a chance, and his impressions are different from what any reliable data or other users will tell you?!

how about some data to back up your "famous i7 lag" claim?

 

[galego]

it must be a problem with his specific PC

According to CHADBOGA, this is a problem that affects to all the members of a given Church that he mentions, although he fails to give a single evidence supporting that ridiculous claim.

how about some data to back up your "famous i7 lag" claim?

Google is very smart and if you write "i7 lag" in the search box, it automatically suggests you 10 different search strings that people uses. Select some of the general ones such as "intel core i7 lag" "core i7 lag" or "i7 lag" to find users reporting similar problems than reported in this same thread.