AMD - Piledriver chitchat

[CAT-THE-FIFTH]

The fact you need (?) a Zxx board to overclock the Pentium dampens its appeal above all for me.

MSI and a few other companies are starting to say they will introduce overclocking motherboards with the B85 chipset(and maybe BIOS updates),so there is hope still unless Intel starts trying to stop it again.

[Xlucine]

The fact you need (?) a Zxx board to overclock the Pentium dampens its appeal above all for me.

With the cost of the motherboard included, there's only a ~£40 difference between a pentium overclocked system and an A10 7700k system. £40 is also about the difference between a R7 250 and 750ti, I wonder how an R7 250 crossfired with the IGP compares to a 750ti driven by that pentium?

[Xlucine]

Is it me,or does TH not actually look at their results again??



Emm,nope??

They even say the Pentium dual core stutters in Thief and the Athlon II X4 does not.

Hang on - the overclocked pentium does have a higher average and minimum fps in every game tested (bar BF4 where it was only the average fps that was higher, but as the graph shows that was most likely some random error in the test since the athlon lost average fps when overclocked), so by the metric most commonly used by all computer review sites it's reasonable to say it performed better.

I also spotted this story on TS:
http://www.tomshardware.co.uk/asus-o...ews-48192.html
H81 asus motherboards can be had for about £40, so there's pretty much no difference in price between an intel pentium system and an athlon system (assuming the cheapest H81 and A88X motherboards will overclock those cpu's as well as the boards used in that review)

[CAT-THE-FIFTH]

Hang on - the overclocked pentium does have a higher average and minimum fps in every game tested (bar BF4 where it was only the average fps that was higher, but as the graph shows that was most likely some random error in the test since the athlon lost average fps when overclocked), so by the metric most commonly used by all computer review sites it's reasonable to say it performed better.

I also spotted this story on TS:
http://www.tomshardware.co.uk/asus-o...ews-48192.html
H81 asus motherboards can be had for about £40, so there's pretty much no difference in price between an intel pentium system and an athlon system (assuming the cheapest H81 and A88X motherboards will overclock those cpu's as well as the boards used in that review)

You need to consider the frametime results too,especially with the Core i3 too.

The stuttering is noticeable in Thief according to Toms Hardware and it was shown in another review of the G3420,that there were noticeable dips with the CPU with the game:
http://oi60.tinypic.com/x37q0j.jpg
http://oi58.tinypic.com/2nivdli.jpg

Chap said the stuttering was noticeable. Look at the GRID2 results also,the frametimes are worse with many largish spikes.

We have already gone through this with graphics card testing,and its relevant to CPU testing too.

Regarding the motherboard situation I have already posted that on another forum,and the earlier noise from MSI over on here:

http://forums.overclockers.co.uk/sho...php?t=18606183
http://forums.hexus.net/pc-hardware-...herboards.html

I did not get around to plonking it over on here - I post on many forums,so there is a delay at times!!

Also,I was talking to one chap over on OcUK(and saw someone do a similar test to him on another forum),and it appears the current lock on H81/B85 motherboards is CPU microcode related. Even Asus,make a disclaimer if you look at my thread on OcUK regarding the new workaround.

[Biscuit]

Holy poop!

http://wccftech.com/ibm-hand-chip-bu...-game-changer/

I wonder what impact this will have on GF?

[CAT-THE-FIFTH]

Holy poop!

http://wccftech.com/ibm-hand-chip-bu...-game-changer/

I wonder what impact this will have on GF?

I read somewhere that already nearly 200 engineers from IBM are working at GF facilities too.

GF has already got into bed with Samsung too.

[DanceswithUnix]

Hang on, going back to that TH benchmark setup, they used a Titan GPU on these??

I know the old theory that you don't want the benchmarks to be bottlenecked on the GPU, but surely the driver has to create enough threads on the gpu to keep it busy, and that must take CPU power doesn't it?

So is pairing a Titan with a low end cpu pointless or actually harmful? I don't remember ever seeing a test that shows it either way. It certainly seems silly, who the hell would ever built that?

[CAT-THE-FIFTH]

I see your point but at some point,it is quite possible that Geforce Titan performance will be seen in a £150 card in say two years time for example. Maybe two sets of benchmarks need to be done,ie,with a cheaper card and a top end one.

I still however would prefer a Core i3 or an FX6300 over a Pentium dual core. You are only talking around a £20 to £25 difference now.

One of the reasons I changed over to a Xeon E3 from a Core i3 2100,was when I noticed that in some games,the Core i3 2100 was starting to become a limitation. Crysis 3 MP was one of those games.

[HalloweenJack]

erm chaps:


https://twitter.com/amd_roy/status/4...845184/photo/1

Something new is coming... @AMDFX

new AM3+ FX???

[watercooled]

My initial thought when I saw FX was an FM2+ chip to coincide with the similar mobile parts. But then I spotted the liquid cooling bit - maybe it's just the 9000 series released with a cooler rather than being sold OEM?

[Xlucine]

I wonder if it is just adding a cooler, or if it includes a die shrink to 28nm? Be nice to see steamroller in there too, but that's too much to expect given the short life before new chipsets and CPU's with DDR4 support.

[AETAaAS]

Much as it goes against what AMD typically does with its socket longevity, I hope there is a total refresh soon. Not just for DDR4 as Xlucine mentions, but I feel that the lack of new mobos particularly mITX is partially due to manufacturers reluctant to put something out for an aging part. Intel is like a teenage girl in this regard and partner companies like this. They have something new and shiny to put on the market much more frequently and that indirectly nets Intel more exposure, particularly at shows like Computex just gone. I feel a chart depicting the new Intel vs AMD mobos will be heavily in favour of the blue team.

[scaryjim]

So ... best case scenario it's an entirely new 8-core enthusiast FX chip.

Moderate scenario is that it's a higher clocked Kaveri part given an FX branding to tie in with the new FX mobile chips and a liquid cooling system to handle the extra heat/voltage required to run Kaveri at high clock speeds.

Worst case scenario it's an existing FX 9000 series chip bundled with a cooler, bringing nothing new to the table at all.

Which one do we think is most likely, guys?

[DanceswithUnix]

First guess: The 8350 has been out for an age. The silicon process must have improved, AMD should be able to lob out a slightly faster part that stays within the 125W envelope. A traditional speed bump like we have seen for years.

More hopeful, the Richland FM2 mid life bump was a complete re-tune rather than just a release of a couple of higher speed grades. Perhaps they can squeeze some new tricks out of the old AM3+ dog yet.

Less hopeful, the 9000 series sold so badly they have crates full of liquid cooling systems in a warehouse somewhere. Slap one on a Kaveri, and print FX on the box.

I notice some new parts have been getting listed as available recently, like http://www.scan.co.uk/products/amd-a...mhz-44w-retail & http://www.scan.co.uk/products/amd-a...mhz-65w-retail so perhaps something is afoot.

[mikeo01]

AIO with 9590. Posted over at OCN

[Xlucine]

Makes sense, they can start marketing the "first stock 5GHz CPU!!1!" to consumers