AMD plans to launch two new Piledriver CPU's.

[gnusounduave]

I can't say I'm really shocked by the news; AMD did this with bulldozer but news is news.

http://blogs.amd.com/fusion/2013/04...ine-welcomes-two-new-additions-to-the-lineup/

 

[storm-chaser]

Yes indeed, I've got my eye on the FX-6350. Looks like pretty good bang for the buck.

 

[trents]

The 4350 has raised the TDP rating of the 95W 4300 to 125W. I'm little surprised at that. Have we gained anything? Or maybe the 4300 should have been rated at 125W TDP all along.

 

[RGone]

"trents" when they came with the 4170 after the 4100 it jumped to 125W TDP also. So that is not new or unexpected really. Then if you read what else is claimed to be different about those two, to be released, cpus; you will know they had to up the TDP.

The AMD FX-4350 processor, for enthusiasts who like to watch crisp HD video and run multiple compute-intensive apps, is a 4-core CPU that clocks at a 4.3GHz Max Turbo operating frequency and 12MB of L2 and L3 cache in a 125W thermal envelope. With up to 10% more performance[ii] than our previous generation[iii] and a competitive price of $122,[iv] this is a hard-to-beat part. Previously the 4 core had 6MB of L3 cache. Now I don't know if they are suddenly adding the L2 and the L3 to come up with that number since we never had to do so before. Maybe just some more creative Hype.

Also announced today, the AMD FX-6350 is a 6-core CPU with up to 4.2GHz Max Turbo operating frequency and 14MB of L2 and L3 cache. This 6-core CPU gives PC users capabilities to accomplish breathtaking 3D modeling and HD video editing. This part is all about bringing maximum multitasking with up to 10% more performance[v] than our previous generation[vi], all for the great price of $132[vii]. Previously the 6 core had 8MB of L3 cache. Now I don't know if they are suddenly adding the L2 and the L3 to come up with that number since we never had to do so before. Maybe just some more creative Hype.

If they truly have increased at least the L3 cache, the TDP would have to rise. There is another consideration that most more proficient than myself will wonder at and that is hits in prefetch that are misses can now take the cpu longer to get the real info to process. With AMD now going to be nearly a year late with Steamroller the hype about 5.0Ghz Centurion and just about anything new they can release to pull people in is no real surprise to me.
RGone...

 

[Woomack]

I don't think that this slow cache is actually helping much. Good to see it has 12-14MB L2+L3 but I don't think that most applications can take any advantage from really large L2 like 6-8MB.
Their weak point is slow access time and generally slow memory ( both cache and memory controller ) while AMD is adding more cache ...
It just still not looking good on a current platform. AMD should find a way to lower generated heat and make access time quicker and they are just adding things that look good on a paper.

 

[trents]

Let's see, 125/95=131% more heat for a 10% increase in performance. RGone, I'm not sure that's a good trade off. In previous generations of AMD CPUs the "tweener" model number CPUs (xx50 as opposed to xx00-Xx00) that came out in the second wave has generally meant better power management, cooler running. I guess I was looking for that sort of change out of the FX-4350 and the FX-6350, from an overclocking standpoint particularly.

 

[Anonaru]

Let's see, 125/95=131% more heat for a 10% increase in performance. RGone, I'm not sure that's a good trade off. In previous generations of AMD CPUs the "tweener" model number CPUs (xx50 as opposed to xx00-Xx00) that came out in the second wave has generally meant better power management, cooler running. I guess I was looking for that sort of change out of the FX-4350 and the FX-6350, from an overclocking standpoint particularly.

This is what I was looking at. Also, based on the "calculation" for 14MB L2 and L3 cache, the two chips (6300 and 6350) have the same number.. So I assume cache didn't change.

I'm a little irked that this isn't a revision or something that addressed some power usage concerns in the process.. Really just looks like they made themselves a new bin and amped up the frequency a bit. Come on now, AMD

 

[RGone]

Now this makes a little bit more sense. Well not sense in that the cpus are getting better but more sense than the first blurb about the two new cpus made to me.

AMD Launches the FX-4350 and FX-6350 Piledriver CPUs
The latest additions to AMD's FX series of CPUs are essentially faster and hungrier versions of the FX-4300 and FX-6300 processors.

 

[dostov]

Yep, they're just "factory OC'ed" versions, hence the higher TDP.

By the way, AMD also slashed prices on the parts they replaced:

The FX-4300 now sells for $108, down from $122, while the FX-6300 goes for $112, down from $132. The FX-8320 is the priciest chip to get a haircut. AMD slashed its price from $169 to $153.

http://www.fudzilla.com/home/item/31250-amd-slashes-desktop-cpu-prices

That FX-8320 is a nice deal at 153$!

 

[RGone]

Just a little heads up about a pattern that seems to be surfacing about the PileDriver/Vishera series.

Back in a time past we bought the cheaper cpu because it was an underclocked version of a faster cpu and then overclocked the less costly cpu to the same speed as the more costly cpu. Most often the less costly processor would reach the same speed as the more costly cpu when overvolted and overclocked.

Well today it seems that AMD actually knows exactly "what" the more costly cpu is and it appears that only the more costly cpu really will do the best. We are now seeing a lesser FX-6300 that will not begin to clock as high as FX-8350. Older wisdom would have us believe the FX-6300 was just an 8 core that had failed cores and would still clock up very well. So far (from a limited sampling for sure) that does not seem to be the situation anymore. You buy a top of the line FX-8350 and mostly it will clock well and do as well as these 8 core cpus from AMD can do. But now when you buy the lesser AMD FX-Piledriver, you get less overall when you then try to match the speed of an FX-8350.

This limited sampling seems to indicate that the Vishera cpus with less cores are not just disabled core cpus but are truly lesser cpus overall.

Now will the two newer cpus (FX-4350/6350) entering the market be the type of cpu we used to buy to overclock? Have there have been cpus that were held aside for later release because they do have less than 8 cores BUT will overclock similar to the FX-8350? Only time will tell. Is AMD at the point in testing/binning their cpus where a lesser cpu is really a lesser cpu overall? That seems the pattern to date with the Vishera series of processor.

It seemed that you could buy an FX-8120 and overclock it to about the same speed as an FX-8150 would go to when clocked wide open, but that does not seem the situation if you buy an FX-8120 and try to push it to the same speed of an overclocked FX-8350. Something seems to have changed. FYI.
RGone...

 

[dostov]

I agree with you, they must use more advanced binning.

FYI, the "new" FX-4350 and 6350 have the same L3 cache as an FX-83xx, pointing to the fact that they are 8-cores parts with disabled modules, but not disabled cache. While I'm not sure what kind of performance increases comes from the extra cache, we know for sure it brings higher power consumption.

 

[RGone]

While I'm not sure what kind of performance increases comes from the extra cache, we know for sure it brings higher power consumption.

Yes it surely will be more power hungry. That does not bode well for those that are buying the lesser boards. Plenty of problems with lesser boards and now it is possible to have even greater issues with lesser VRM circuits.
RGone...

 

[Nebulous]

What I don't like is this:

"AMD’s product warranty does not cover damages caused by overclocking, even when overclocking is enabled via AMD software."

Yet with intel you can purchase insurance for such an event.

I do agree that AMD is cheaper so I guess that's a plus.

 

[trents]

RGone, I think that's the pattern I've been seeing in following threads on this forum by FX-43xx and 63xx overclockers. I agree with your observation. They are inferior in more than just having cores locked.

 

[gsxrme]

The new 6 core should overclock higher than the 8 cores and how many apps or games really use those 8 threads? I think its a good launch for AMD

 

[RGone]

The new 6 core should overclock higher than the 8 cores

It is apparent that you have not read what has been written in this thread at all. Mostly the exact opposite of what you said is what we are now seeing a pattern about. The lesser cores do n0t seem to overclock better than the 8 core cpus.
RGone...

 

[rescuetoaster]

What I don't like is this:

"AMD’s product warranty does not cover damages caused by overclocking, even when overclocking is enabled via AMD software."

Yet with intel you can purchase insurance for such an event.

I do agree that AMD is cheaper so I guess that's a plus.

To my knowledge, no one has ever covered overclocking. In fact, it is almost always listed as a specific reason why your RMA request could be rejected. I haven't heard of this "insurance" thing through intel, but that could be one way AMD could make a few extra bucks - just sell extended warranties that cover overclocking.

 

[trents]

To my knowledge, no one has ever covered overclocking. In fact, it is almost always listed as a specific reason why your RMA request could be rejected. I haven't heard of this "insurance" thing through intel, but that could be one way AMD could make a few extra bucks - just sell extended warranties that cover overclocking.

Does "Don't ask, don't tell" apply in the case of AMD CPUs fried from overclocking?

 

[Anonaru]

Does "Don't ask, don't tell" apply in the case of AMD CPUs fried from overclocking?

As far as I know, yes. They don't have a very reliable way of really finding out, and really it isn't worth the effort. I blew up a 6300 on my LN2 table, and I'm fairly confident I could get that replaced if I wanted to. (Of course, I'm being a good boy, disassembling the part as much as I can, and sticking it into a cool fraken exhibit of some kind in the shop

I'm not very excited at all regarding these new parts. That Cache sounds like the same old marketing crap (L2+L3 Cache combined total, hence why the two parts have different numbers), though it would be nice if this isn't the case..

A slightly higher clock pushed them above a 120w power envelope. A 4-core, mid / low end current generation part (Priced as such) that requires a premium motherboard to do anything with. Great

I was hoping they would employ those Richland-style optimizations to Vishera, and with those get these same factory overclocks without blowing the TDP into the stratosphere, but it really sounds like they just nixed some cores or got fancy bins and just bumped up the clocks. Bad AMD bad!

 

[storm-chaser]

It is apparent that you have not read what has been written in this thread at all. Mostly the exact opposite of what you said is what we are now seeing a pattern about. The lesser cores do n0t seem to overclock better than the 8 core cpus.
RGone...

Sounds like you are thinking more heat and more stress on the vrm circuits due to the higher power requirements? Makes sense, but we don't really have any tangible benchmark results on these new releases yet.