Vishera Review Up - Anandtech

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Olikan

Platinum Member
Sep 23, 2011
2,023
275
126
Of course it has the CMT perf. hit,it has the same shared decoder and same shared (single) FP unit. That's why they claim 80% performance of "traditional" dual core design in less die area. Sharing of the frontend and FP unit makes it less efficient versus designs with dedicated execution units(like Phenom/Core). But looking at the MTed performance those 4 FP units can now significantly outperform Thuban's 6 cores each of which has a full fp unit to itself. Problem is that still each integer core cannot fully utilize the whole FP unit when there is lightly threaded workload and Win7 scheduler didn't help. I wonder how much if at all Win8 will improve things.

i thought that we would se something diferent than 20%, piledriver fixed many small things, and the "decoder/cache/f-point" bottleneck (aka big ones) could have a diferent hit on performance
but nop, straight 20%...
 

inf64

Diamond Member
Mar 11, 2011
3,884
4,691
136
That's why they made SR core . They ditched the shared decoder and they optimized the schedulers. Bad news is that FX based on SR core will most likely not launch next year.

PD does have better IPC even with the same hit due to CMT,all those small improvements add up.

You can see how close to Llano Vishera(PiledriverV2) really is. On average just 10% less IPC than Llano. It does have the CMT loss when you run it like 2M/4C though so best case is Win8+8320/8350.
 
Last edited:

pelov

Diamond Member
Dec 6, 2011
3,510
6
0




Still ~20%. That's not going to be addressed until they widen the front end in Steamroller.

Does anyone know if Steamroller with have dynamic cache values? As in if tasked with only a single thread on a module, it has access to all of the L2? I recall them stating something like quarter block sizes? Where they can power down the cache in 25% increments
 

inf64

Diamond Member
Mar 11, 2011
3,884
4,691
136
@ Pelov
Yes, it should have dynamically realizable L2 cache. It's more of a power saving technique since PD's single core within a module can use up the whole L2 if possible. What they will need to do is improve L2 read and write performance and this is what they claim they achieved(especially write). OoO loads and store are more efficient also(supposedly).
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
woot, found something interesting...piledriver have the same problem as bulldozer, 20% ipc hit in CMT
http://www.extremetech.com/computin...ledriver-deliver-where-bulldozer-fell-short/2




Uh, that's not so much a problem as part of the choice of using a CMT design. Providing more thread processing per mm squared than a conventional core design. In fact that was the one thing that Bulldozer didn't underperform on, it met or exceeded the pre-launch estimates in regards to CMT.

A more CMT aware OS helps avoid non-optimal loading, which I believe is one of the reasons Linux comparisons of Bulldozer weren't quite as terrible as on Windows 7.

Wonder if HT could work well with an Intel CMT design...
 
Last edited:

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
Wow 200 bucks for a 8 core CPU I mean what a nice deal. I wonder how they OC anandreview didnt say that.. what 6Ghz 8 cores or ivy 3960x 6 core,,? which one,,

I take it the 12 thread intel monster will beat it to a help. Wasnt Vishera a wrestler in WWF long time ago. gl
 

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
@ Pelov
Yes, it should have dynamically realizable L2 cache. It's more of a power saving technique since PD's single core within a module can use up the whole L2 if possible. What they will need to do is improve L2 read and write performance and this is what they claim they achieved(especially write). OoO loads and store are more efficient also(supposedly).

Wouldn't that introduce problems too? Say, bumping up another thread's stores to LLC which isn't going to get any better, judging from AMD's last statements regarding Steamroller? For power consumption that would do wonders, though. With that much cache, you need a way of toggling that on and off.

The read and writes have improved, although still lag behind Intel.





IMO, this should be AMD's #1 concern: cache and IMC. Cache is still the most likely reason for the PD architecture underperforming in lightly-threaded workloads and the IMC is going to be their biggest bottleneck in their APUs. They've made improvements, even if the L2's cycles don't seem to have changed, but not enough to notable.

A more CMT aware OS helps avoid non-optimal loading, which I believe is one of the reasons Linux comparisons of Bulldozer weren't quite as terrible as on Windows 7.

AMD advise against loading modules first then doubling up on threads after you've hit them all. The power consumption while doing that would be much worse than it already is. The win8 scheduler doesn't do that either. The scheduler is much more complicated, taking into account like threads sharing resources, turbo, and more.

The two biggest reasons BD/PD do better in Linux are 1 - Linux tends to have fairly well threaded software, and 2 - the GCC compiler.
 
Last edited:

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
Bit more CMT info



I believe Fritz is very light on cache, so the graph above shows the efficiency of SMT vs. CMT without the burden of the L2/L3, especially given the sacrifices AMD had to concede in order to make such a chip. It definitely closes the gap, but only after 4 and the closer to 8 the better.
 

formulav8

Diamond Member
Sep 18, 2000
7,004
522
126
The 8320 will be my new cpu it seems. Its well under $200 and is not near as stinking boring as the Intel competitors. And will still give me great bang over my current Intel based Core2q. Around 4.8-5ghz would be the sweet spot.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
Woah, calm down man.

I am calm . Its not me .People are so threatened AMD is going out of business that they look at these underperforming power eaters and say not bad . Its amusing . This is amds best 8 core cpu . There in deep do do. Comparing to intells 4 cores is fun and all based on price . But what is exactly the reason for a price compare to beginn with . Because WHY exactly? It has nothing to do with price. A 6 core intel wipes the floor with these . You guys might think based on PRICE Amd has a nice cpu. Thats laughable. I understand the price compare is good .. Same as cpu efficiency is a big deal . Bottom line Intel is cheaper to own threw the lifetime of the product by a large factor . But than The same people who sing the praises of cheaper cpu . Avoid the fact intel 8threaded 4 core is cheaper to own and a better bargin. AMD competes with the 3750 so why even show the 8 thread model in the reviews . In so doing the reality is hidden even deeper in this tangled story of Intel Vs AMD and confussing the cost of ownership . If fans think this will save AMD they are wrong . Next QT. tells the story . I just hate the fact I will have to listen to 3 more months of How great this CPU and the earnings report will show something all together differant . Stange world
 

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,205
126
I am calm . Its not me .

You guys might think based on PRICE Amd has a nice cpu. Thats laughable.

Stange world

Well, most people DO compare based on PRICE. It's probably the largest part of the equation.
It's completely normal. Your pro-Intel rants, on the other hand...
 

hokies83

Senior member
Oct 3, 2010
837
2
76
Looks like the 8350 competes with the i7 950... and keep in mind the i7 950 is at 3ghz vs the 4.2ghz 8350 and all of us who had 1366 or still do know you can get 4.5 = 4.8ghz pretty easy out of them.

Im sure if we clock that 950 up to 4.2ghz we would see close results. You can pick up those combos for 200$ all day long..

http://www.anandtech.com/bench/Product/100?vs=697

Gaming wise it Competes with the i5 760.. which is clocked at a mere 2.8ghz..

I myself own 2 of these and paid 130$ for the MB=Cpu...

So i see a fail in price per performance big time here.

http://www.anandtech.com/bench/Product/191?vs=697
 
Last edited:

CHADBOGA

Platinum Member
Mar 31, 2009
2,135
832
136
Looks to me like AMD just created something worth buying again. All the reviewers love to point out that AMD doesn't make anything to compete with the high end, however average Joe buying a motherboard and processor isn't going to spend more than $199 on his processor and right now AMD has a nice option at that price point and lower. Plus who wouldn't love to tell their buddies that they managed to OC their proc. to 5GHZ OMG!!!

I like how "average Joe", wants to overclock to 5Ghz(so he has at least that kind of knowledge), but still doesn't give a damn about single core performance(with that being more relevant to most desktop users). D:
 

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
When will Johan get his hands on one of these? or the 8-module MCM parts? I don't think AMD cares much how it does on the desktop as the server was its intended purpose in the first place.
 

formulav8

Diamond Member
Sep 18, 2000
7,004
522
126
Desktops ='s going the way of the dodo. I, like others here will always prefer a desktop over anything else. While there will always be some kind of desktop market, it will end up being one of the least important, Literally.

So while I personally wouldn't blame AMD for focusing less on us, plus some others, I don't like the thought of fewer choices. Intel already forces us to spend over $200 to have any kind of real cpu fun. They already have the most boring cpu's when it comes to reduced budget tweakers who like the tweaking game.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
Says the guy with the GTX 470's.()

At least it's a decent improvement. Just hope they survive with this...

Waiting for GK110, as painful as it's been recently I've overclocked these cards to the moon and back can't wait for something new. :\

Yes it's nice to see them moving forward instead of backwards. It's just a pity they're so far behind so much ground to make up so little cash to make it happen

I'll probably be upgrading next year, hopefully they have something else out. At least they learned from their initial hype train/awful pricing - major improvement in many aspects over the bulldozer release.
 

Chiropteran

Diamond Member
Nov 14, 2003
9,811
110
106
I'm actually glad they didn't puss out and try to reduce power consumption. I'd rather have the performance than save what amounts to pennies a year on the electricity bill.
 

Agent11

Diamond Member
Jan 22, 2006
3,535
1
0
They need to leap to 22nm as soon as possible to even hope to compete on power levels.
 

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
They need to leap to 22nm as soon as possible to even hope to compete on power levels.

32 to 22 isn't going to knock off half their consumption, they need a better design.

Bulldozer was one step backwards, Vishera is two steps forward. They have a ways to go still.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |