Whos getting Vishera/new FX lineup?

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

inf64

Diamond Member
Mar 11, 2011
3,884
4,692
136
IPC is 5% better? Facts from hardware.fr in a nice chart here.
Both at 4Ghz,turbo disabled. Application workloads average out at 7.7% increase while games average out at 13.5% increase. Just FYI ;).
Oh and it clocks higher than 8150 at stock but has the same max.(single core) Turbo and FX8150 has a very high half core Turbo of 3.9Ghz,100Mhz shy of the 4Ghz Vishera runs at. That's where IPC improvements kick in and that's why Anand can write this in his conclusion :
Anand said:
Final Words

Vishera is a step in the right direction for AMD, it manages to deliver tangibly better performance than last year's disappointing FX processor without increasing power consumption. Thanks to architectural and frequency improvements, AMD delivers up to 20% better performance than last year's FX-8150 for a lower launch price, while remaining within the same thermal envelop

Another confirmation of ~10% IPC uplift with BD->PD (which is average from apps and games Hardware.fr got) is this review .

As can be see, in the chart are Deneb,Llano,BD,Trinity and Vishera -all running at the same 3Ghz clock and having 4 cores or 2M in case of BD derivatives. The workloads are various,there are gaming workloads,encoding,apache server etc.
Deneb has 7% higher IPC than Vishera, Llano is 10% better than Vishera while Vishera is 10% better than Bulldozer and 5% better than Trinity. All lines up very well with Hardware.fr findings.
 
Last edited:

The Alias

Senior member
Aug 22, 2012
646
58
91
IPC on this is 5% higher than Bulldozer and clocks 5-10% higher. Really not sure how some people are excited for this when an i5 is overall faster and consumes half the power.

This is nothing but a small revision of Bulldozer, which is why you only see a ~10% performance increase. How are people going crazy over a 10% increase to a POS?

time%20to%20completion.png

maybe because it outperforms intel at the same pricepoint and higher ?

This is something for AMD to be proud of. Its FX-8350 finishes in second place (of the CPUs I charted—I tried to pick and choose carefully to keep the graph from getting too hectic). For the record, though, I also had to know how FX-8350 did against Core i5-3570K, and it finished 12 seconds before the pricier Intel chip.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
time%20to%20completion.png

maybe because it outperforms intel at the same pricepoint and higher ?

Outperforms an i5 in multi-threaded only, that is. Throw anything there that isn't multi-threaded or is a mix of workloads and it'll be significantly slower. Also, the i5 consumes half the amount of power.

This CPU is irrelevant for most people.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
energy%20used.png


4ghz%20itunes.png


It's still an inefficient POS. Not only will AMD's IGP advantage with Trinity be null when Haswell comes out, but because of the CPU performance upgrades we'll be back to square zero: Sandy Bridge vs. Bulldozer again.

It's good in multi-threaded programs, but consumes far too much power to give you that performance.
 

N4g4rok

Senior member
Sep 21, 2011
285
0
0
This CPU is irrelevant for most people.

I think the excitement extends mostly from meeting expectations. The fact that they were able to produce something that performs better than BD and trades blows in multi-threading with Ivy Bridge is great news considering the company's bleak outlook.

In other words, they're learning. This combined with recent driver improvements shows that they're putting some effort into it again.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
I think the excitement extends mostly from meeting expectations. The fact that they were able to produce something that performs better than BD and trades blows in multi-threading with Ivy Bridge is great news considering the company's bleak outlook.

In other words, they're learning. This combined with recent driver improvements shows that they're putting some effort into it again.

Well, they better be, considering how terrible BD was. Upgrading a POS by 10-15% isn't really exciting, IMO.

It's still a pig in anything that isn't multi-threaded and consumes too much power, especially if you overclock.
 

inf64

Diamond Member
Mar 11, 2011
3,884
4,692
136
What Alias posted (THG chart) is the complete benchmark suite runtime. That means ST and MT workloads combined.Again LolWut fails at understanding,why I'm not surprised. And fails at replying to my posts which prove his IPC claim is wrong ;).
 

N4g4rok

Senior member
Sep 21, 2011
285
0
0
Well, they better be, considering how terrible BD was. Upgrading a POS by 10-15% isn't really exciting, IMO.

No, it's not that exciting. It will likely be met with cautious optimism from the majority of the informed market.

It's still a pig in anything that isn't multi-threaded and consumes too much power, especially if you overclock.

I think there's still a large enough group of buyers/system builders who are not at all concerned with power consumption. Pairing that with the fact that the single threaded performance might be high enough to please those looking for a higher core count, it will probably sell more units than it deserves. Personally, I think that's fine as long as they put that money to good use.
 

jhansman

Platinum Member
Feb 5, 2004
2,768
29
91
Face it folks (and I am a long-time AMD supporter), AMD will likely never catch up with Intel, so stop waiting for them to. This processor looks like a good next step from my X6 1090T, and with a BIOS update, my mobo will run it, so I just might go with the 8350 and call it a day. Not a gamer, Photoshop is the heaviest thing I throw at my CPU and graphics, so what the heck. All this "whose d*ck is biggest" nonsense is for children and gaming geeks who still[\I] can't find their way out of their parent's basement. Kudos to AMD for righting their BD wrongs and hanging in.
 

The Alias

Senior member
Aug 22, 2012
646
58
91
What Alias posted (THG chart) is the complete benchmark suite runtime. That means ST and MT workloads combined.Again LolWut fails at understanding,why I'm not surprised. And fails at replying to my posts which prove his IPC claim is wrong ;).

this . I wonder what he's gonna say when I say the i3s are now irrelevant for gaming because of fx ... oh well

51123.png


51124.png


http://nl.hardware.info/reviews/3311...clancys-hawx-2

http://nl.hardware.info/reviews/3311...rks-cpu-dirt-3

http://nl.hardware.info/reviews/3311...s-cpu-crysis-2

read em and weep buster ! lol
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
No, it's not that exciting. It will likely be met with cautious optimism from the majority of the informed market.



I think there's still a large enough group of buyers/system builders who are not at all concerned with power consumption. Pairing that with the fact that the single threaded performance might be high enough to please those looking for a higher core count, it will probably sell more units than it deserves. Personally, I think that's fine as long as they put that money to good use.

It's not like in terms of power consumption it's a negligible difference. It consumes twice the amount of power as an i5. That means that not only do you need to invest more into fans, but also into a better power supply. The huge power consumption difference alone negates any advantage from a cheaper price, not that there is one because right now it's at $220 on Newegg and the superior 3570K will run you $230.
 

Olikan

Platinum Member
Sep 23, 2011
2,023
275
126
It's not like in terms of power consumption it's a negligible difference. It consumes twice the amount of power as an i5. That means that not only do you need to invest more into fans, but also into a better power supply. The huge power consumption difference alone negates any advantage from a cheaper price, not that there is one because right now it's at $220 on Newegg and the superior 3570K will run you $230.

50% more, not twice :awe:
 

The Alias

Senior member
Aug 22, 2012
646
58
91
Actually, 60% more if you want to be picky. That is still a lot when the performance is lower except in heavily multithreaded workloads.
note that more and more programs are written to be multithreaded hence the reason the 8350 beat the 3570k in time to completion on tom's review
 

inf64

Diamond Member
Mar 11, 2011
3,884
4,692
136
Yes but only under full load and only if you run it during prolonged period of time each and every day(say 10 hours a day,every day in a week and then more than a year like this). Lightly threaded loads and idle will show no such difference. Games for instance show ~20W more (at the wall) for 8350 vs i5/i7 since CPU is not taxed 100% and most of the power is drawn by other components.
 

DigitalWolf

Member
Feb 3, 2001
108
0
0
Personally I am going to get one for my AMD box. I am simply waiting for them to show up on Amazon so that I can use a gift certificate I have.
 

KompuKare

Golden Member
Jul 28, 2009
1,228
1,597
136
k_reltel.jpg
[/URL]
As can be see, in the chart are Deneb,Llano,BD,Trinity and Vishera -all running at the same 3Ghz clock and having 4 cores or 2M in case of BD derivatives. The workloads are various,there are gaming workloads,encoding,apache server etc.
Deneb has 7% higher IPC than Vishera, Llano is 10% better than Vishera while Vishera is 10% better than Bulldozer and 5% better than Trinity. All lines up very well with Hardware.fr findings.

Quick buy a Llano before they're all gone! Is the 3870K or the 3670K better value?

Seriously, FM1 and FM2 both have good idle power mostly better than SB or IB (load's a different story of course, but my computer spends most of its day idling). If only AMD can manage to figure out a way to get ZeroCore Power to work with their APU, I'd be tempted to get one and upgrade to a GPU which supports it. Low idle power plus a GPU which turns itself 'off' until needed would be nice.
 
Dec 30, 2004
12,553
2
76
Actually, the turbo core was one of the few good aspects of the entire architecture. It was snappier and more responsive than Intel's turbo in SB.

I've got a 955 and I'm not planning on upgrading until I need to. This chip will last me another 2 years easy :/
I thought we had some threads that talked about this and it sucks...I'll have to go find them... it's not intelligent enough to put threads that need high performance on the turbo'd cores, among other things...

And who says that Anand's results are right?

His AMD results have been consistently, for year now, lower than the other sites. Furthermore, doing a run on your own you will find you are scoring higher than him. But if you want a better known reviewer than what was linked, read TR's review or Tom's.

I always thought his AMD results were the "fairest" of the bunch, if even a little kind to AMD-- particularly with Cinebench 10, I never could get the score he did clock/clock .
 

Ajay

Lifer
Jan 8, 2001
16,094
8,114
136
As per usual, a "who's going to get one" thread turns into and Intel/AMD fanboi brawl :(
 

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
I thought we had some threads that talked about this and it sucks...I'll have to go find them... it's not intelligent enough to put threads that need high performance on the turbo'd cores, among other things...

We have, but the gist of it this:

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.

You can avoid the ~20% penalty, but you're going to pay in clock speeds, power consumption and perhaps even in performance not being optimal.

I still hate CMT. Unless their HSA agenda succeeds and we see GPGPU acceleration for a substantial amount of applications or we see a drastic change in software parallelization, I'm going to hate it.

I like how Tom's approaches their benchmarks, though. They're only using the synthetic stuff in order to determine some quirks of the architecture, with the majority of the benchmarks being actual applications. I'd much rather know how a chip performs in Handbrake than Cinebench, because only one of those has any use to me :p