AMD Athlon X4 860K review @ pcpop

[Shehriazad]

I will have to link myself in to this discussion here.


The 860K overclocks better than the 7850K...why? Because it actually sports far lower voltages.


I own both 7850K and 860K and was quite surprised.

The 7850K is stock volted at: 1.3525 VID and 1.15 NB VID
While the 860K ist volted at: 1.3250 and 1.075 NB VID


Those are quite the big difference in stock voltage.

I could push my 7850K to 4.4 GHZ stable (no 4.5 QQ) but was forced to raise the CPU VID to 1.43 and NB VID to 1.2 even though the iGPU was disabled.

Meanwhile my 860K sits at 4.8 GHZ stable (endured 10 hours of tests so far, I'll go ahead and claim that it's stable) with a CPU VID of 1.41 and a NB VID of 1.075 (did not have to alter it).

Since the iGPU unit of the 860K is disabled on more than just a software level (it seems) the overall power draw from the NB VID is lower and the CPU is overall 5° cooler in stock.


Those results are all in the same system. Using the same cooler (A measly Scythe Katana, as this is one of the few that fits in my ITX cube).


Once tomorrows' 24 hour stress test was successful I will post a few screens. But it seems I haven't hit a temp ceiling...it just won't boot stable at 4.9 GHZ. I could go for FSB OCing...but I don't like messing with that as too many things are affected by it.


P.s. I'm not sure where any review site gets 1.45 to 1.5 stock volt from...their stuff must be hella broken.

 

[DrMrLordX]

4.8 ghz? Nice! You might have a golden chip, mind, but that is still impressive. I've only seen a few results from the 860k, and that's the first one above 4.6 ghz I've seen on sane vcore.

And yeah 1.5v VID seems just silly.

 

[NTMBK]

I will have to link myself in to this discussion here.


The 860K overclocks better than the 7850K...why? Because it actually sports far lower voltages.


I own both 7850K and 860K and was quite surprised.

The 7850K is stock volted at: 1.3525 VID and 1.15 NB VID
While the 860K ist volted at: 1.3250 and 1.075 NB VID


Those are quite the big difference in stock voltage.

I could push my 7850K to 4.4 GHZ stable (no 4.5 QQ) but was forced to raise the CPU VID to 1.43 and NB VID to 1.2 even though the iGPU was disabled.

Meanwhile my 860K sits at 4.8 GHZ stable (endured 10 hours of tests so far, I'll go ahead and claim that it's stable) with a CPU VID of 1.41 and a NB VID of 1.075 (did not have to alter it).

Since the iGPU unit of the 860K is disabled on more than just a software level (it seems) the overall power draw from the NB VID is lower and the CPU is overall 5° cooler in stock.


Those results are all in the same system. Using the same cooler (A measly Scythe Katana, as this is one of the few that fits in my ITX cube).


Once tomorrows' 24 hour stress test was successful I will post a few screens. But it seems I haven't hit a temp ceiling...it just won't boot stable at 4.9 GHZ. I could go for FSB OCing...but I don't like messing with that as too many things are affected by it.


P.s. I'm not sure where any review site gets 1.45 to 1.5 stock volt from...their stuff must be hella broken.

Interesting, thanks! Perhaps your 860k is from a newer stepping/revision- the 28nm process may have improved over the past 6 months. When did you get your 7850k?

That's an impressive overclock. Sounds like a great setup you have there.

 

[Kalessian]

If you could, would you mind running the dolphin benchmark?

https://forums.dolphin-emu.org/Thread-new-dolphin-cpu-benchmark-no-game-required

 

[Shehriazad]

Interesting, thanks! Perhaps your 860k is from a newer stepping/revision- the 28nm process may have improved over the past 6 months. When did you get your 7850k?

That's an impressive overclock. Sounds like a great setup you have there.

My 7850K was from the very first batch, I got it in the first week of release.
The 860K arrived from a new batch just last friday.


To the other guy with the benchmark program. Will do once I'm back home on friday. (currently away from home because of work..shhhh don't tell my boss I'm on here ;p) I couldn't finish the 24 hour stress test yet either since I had to leave a day early...work sucks!



For people who are interested in the full setup:

Board: Asrock FM2A88X ITX+
CPU: 860K @ 4.8 GHZ
Ram: Team Vulcan 2400 MHZ 16GB kit. (Since the board supports 2400 MHZ out of the box there was absolutely no problem just running it at 2400)
GPU: Palit 750 Ti StormX OC
OS: Win 8.1 64Bit
PSU:Be Quiet Pure Power L8 (350W) 80+ Bronze
OS SSD = Sandisk 64 GB
Game SSD = m.Sata Crucial M500 240GB (Was quite surprised that the Asrock board actually had an m.Sata slot...only ITX board to have one for FM2+)
Case: Cooltek U2 -> meh airflow, but the PC remains very portable...so it got that going for it...which is nice.

 

[crashtech]

My 7850K was from the very first batch, I got it in the first week of release.
The 860K arrived from a new batch just last friday.


To the other guy with the benchmark program. Will do once I'm back home on friday. (currently away from home because of work..shhhh don't tell my boss I'm on here ;p) I couldn't finish the 24 hour stress test yet either since I had to leave a day early...work sucks!



For people who are interested in the full setup:

Board: Asrock FM2A88X ITX+
CPU: 860K @ 4.8 GHZ
Ram: Team Vulcan 2400 MHZ 16GB kit. (Since the board supports 2400 MHZ out of the box there was absolutely no problem just running it at 2400)
GPU: Palit 750 Ti StormX OC
OS: Win 8.1 64Bit
PSU:Be Quiet Pure Power L8 (350W) 80+ Bronze
OS SSD = Sandisk 64 GB
Game SSD = m.Sata Crucial M500 240GB (Was quite surprised that the Asrock board actually had an m.Sata slot...only ITX board to have one for FM2+)
Case: Cooltek U2 -> meh airflow, but the PC remains very portable...so it got that going for it...which is nice.

Since you seem willing to take requests, I would love to see your Cinebench 11.5 score posted. CPUs like yours are very underrepresented, I feel it may be a good budget alternative for highly threaded games, etc.

 

[lixlax]

Its quite power efficent oo, compared to the FX line. Under BF4 64p my 860k+R9 270 takes about 240W using mantle and about 220W when using DX11. My previous build of FX8320+HD7850 took 280-290W under CX11(I didn't try mantle on this system), both systems at stock clocks and rest of the system being similar. Not sure though how much more power the 270 is sucking compared to the 7850. Idle power consumtion is also about 15W less than the FX build had.

 

[Spawne32]

http://www.overclock.net/t/1522956/860k-preliminary-results-vs-8320/0_20

results if you guys are interested

 

[Gikaseixas]

http://www.overclock.net/t/1522956/860k-preliminary-results-vs-8320/0_20

results if you guys are interested

Nice showing for that little guy compared to the FX, much better single thread performance

 

[DrMrLordX]

The SuperPi performance is curiously good. I would still rather have an 8310 or 8320E for MT workloads, but if my concentration were gaming, the 860k would be my go-to CPU vs the FX.

 

[SPBHM]

The SuperPi performance is curiously good. I would still rather have an 8310 or 8320E for MT workloads, but if my concentration were gaming, the 860k would be my go-to CPU vs the FX.

gaming should be better with a the 8300s,
some games are using more than 4 threads and the l3 cache seems to help considerably for games

it would be interesting to compare with the 4300s, but I would guess 4300s are easier to OC (?) and the l3 is going to help.

 

[Spawne32]

The SuperPi performance is curiously good. I would still rather have an 8310 or 8320E for MT workloads, but if my concentration were gaming, the 860k would be my go-to CPU vs the FX.

I was surprised by that as well, if ram has any effect on it, im only running DDR3-1600 ballistix elite sticks with the XMP profile enabled. I am looking to upgrade to DDR3-2400 in the near future to see how much more performance I can get out of a memory upgrade.

 

[inf64]

SR core does outclass PD core (with L3) in certain workloads by a not so small margin. One such workload is superpi. There are others but superpi is sort of an outlier.

Anyhow, as much as 860K is a good chip, FX6300 for the price/perf. outclasses it completely. You get 6 threaded unlocked PD core with L3 cache that can OC to around 4.5-4.6Ghz on air. The downside is the AM3+ platform but for most gamers it would be a solid solution ( Asus M5A97 LE R2 board is very cheap and supports 140W SKUs, while maintaining great build quality).

 

[SlowSpyder]

The SuperPi performance is curiously good. I would still rather have an 8310 or 8320E for MT workloads, but if my concentration were gaming, the 860k would be my go-to CPU vs the FX.

Yea, that's pretty good. I have to go above 5GHz to get low 17 second scores on one thread. I think my best was something like 16.9 seconds at over 5.3GHz on my FX. As I understand it, SuperPi performance means nothing as far as real world performance goes, but still interesting for comparing architecture to architecture.

 

[DrMrLordX]

gaming should be better with a the 8300s,
some games are using more than 4 threads and the l3 cache seems to help considerably for games
it would be interesting to compare with the 4300s, but I would guess 4300s are easier to OC (?) and the l3 is going to help.

I dunno, the L3 on piledriver is pretty slow. Not very many games are using more than four threads, either. There are some, but . . .

I was surprised by that as well, if ram has any effect on it, im only running DDR3-1600 ballistix elite sticks with the XMP profile enabled. I am looking to upgrade to DDR3-2400 in the near future to see how much more performance I can get out of a memory upgrade.

When it comes to Pi, anything that reduces your memory latency is going to offer you an improvement. Try clocking that DDR3-2400 down to DDR3-1800 or so and see how low you can get your timings.

SR core does outclass PD core (with L3) in certain workloads by a not so small margin. One such workload is superpi. There are others but superpi is sort of an outlier.

Anyhow, as much as 860K is a good chip, FX6300 for the price/perf. outclasses it completely. You get 6 threaded unlocked PD core with L3 cache that can OC to around 4.5-4.6Ghz on air. The downside is the AM3+ platform but for most gamers it would be a solid solution ( Asus M5A97 LE R2 board is very cheap and supports 140W SKUs, while maintaining great build quality).

Honestly, I wouldn't want a 6300 unless I knew for sure I didn't need 8 threads and I wanted to save maybe $30 on the chip. The new 95w octal-core FX chips made the 6300 far less appealing.

Yea, that's pretty good. I have to go above 5GHz to get low 17 second scores on one thread. I think my best was something like 16.9 seconds at over 5.3GHz on my FX. As I understand it, SuperPi performance means nothing as far as real world performance goes, but still interesting for comparing architecture to architecture.

Pi doesn't count for much. It's only really useful for benching a system against itself. We can infer that Steamroller handles SSE2 better (assuming that's SuperPi Mod 1.5 XS), but not much else. It also exposes the slow L3 on Piledriver for what it is.

 

[FlanK3r]

Wow, I would like to see 4.8 GHz 860K....Im looking for some results here man.

 

[Spawne32]

Wow, I would like to see 4.8 GHz 860K....Im looking for some results here man.

Unfortunately highest im able to go safely is 4.6ghz, I presumed that the MSI A88XM-Gaming board had 6+2 phase and a decent overclocking bios like it was advertised at CES but I was wrong. Without the ability to adjust LLC im bumping 1.6vcore just to get 4.6ghz stable with 1.532v under load. The nepton 140xl can handle more but im worried about pushing the voltage any higher.

 

[DrMrLordX]

We'll just have to wait on some results from Shehriazad.

 

[Spawne32]

We'll just have to wait on some results from Shehriazad.

I'm sure there will be plenty more results to come, apparently only a small amount shipped from the release date, my first chip was DOA, and I had a hard time getting the second chip for the same price I got the first, most places were out of stock and the only store that had it close to the 83 dollar price I bought it at off amazon was frys. Right now I have the third highest frequency on hwbot, so I would imagine not that many have gotten into the hands of overclockers.

 

[DrMrLordX]

I'm sure there will be plenty more results to come, apparently only a small amount shipped from the release date, my first chip was DOA, and I had a hard time getting the second chip for the same price I got the first, most places were out of stock and the only store that had it close to the 83 dollar price I bought it at off amazon was frys. Right now I have the third highest frequency on hwbot, so I would imagine not that many have gotten into the hands of overclockers.

Let's hope so. The possibility that the 860k is the first of a line of improved 28nm planar chips from AMD is somewhat-encouraging.

On a side note, have you tried NB overclocking on your 860k yet? That totally slipped my mind. Kaveri hates NB clocks higher than 2 ghz.

 

[Spawne32]

Let's hope so. The possibility that the 860k is the first of a line of improved 28nm planar chips from AMD is somewhat-encouraging.

On a side note, have you tried NB overclocking on your 860k yet? That totally slipped my mind. Kaveri hates NB clocks higher than 2 ghz.

I did try booting above 2ghz and even with more voltage the board failed to boot. It's possible I could get it to go above 2 with little bit more however. Havent played with it much yet. That is something ill be looking to do this week though.

 

[Spawne32]

Also on a side note, how would you go about testing the stability of the northbridge, OCCT i would imagine?

 

[DrMrLordX]

NB testing is sketchy, but yeah something like OCCT or IBT would work. Just make sure you're maxing out memory usage. Memtest86+ is also a good tool for that sort of testing, as is Prime95 Blend.

 

[Kalessian]

Would you take a shot at the dolphin benchmark I posted earlier?

I'm also interested in NB overclocking. Also, any way to test power consumption? Mostly interested in stock vs highest OC load consumption.

 

[crashtech]

This AMD offering may convince me to do my first AMD build in a long time. More data would be helpful, and to know also what kinds of speeds are achievable with the stock cooling solution.