AMD Trinity to launch on October 2nd

[Rezist]

Needs DDR4 now maybe steamroller?

 

[blckgrffn]

The best thing about this is that FM2 will stick around.

I think trinity will be a very nice processor for mainstream consumers and general workloads as long as they get the higher clocked versions.

Agreed.

*Remembers 2.1 ghz tri-core FM1 Llano, shudders*

 

[Phynaz]

Trinity is a class leading product for sure.

What class would that be? The kindergarten class?

 

[piesquared]

What class would that be? The kindergarten class?

Ahh good, more intel trolls. The more of you that show up trolling AMD threads the better, its a good indication of the threat to intel Trinity really poses.

 

[zaydq]

Sounds exciting. I remember loving my Athlon 64 cpu and ocing it. It'd be great to see AMD out with great cpus again. I've already got an i5, but if they can get their crap together, next build could very well be AMD based.

 

[Yuriman]

I own both AMD and Intel chips. I owned nothing but AMD prior to 2007. I still own only AMD video cards, except in my laptop because NV was unquestionably a better value at the time, and I don't let my fanboyism sell me an inferior product.

I am unimpressed with the FX line, but it's clear how it happened. Intel has superior fabs and many times the funding for R&D. I'd love to donate money to AMD's cause but I'm a poor college student and I don't have money to give.

Truth is, LLano is slower per clock than Kentsfield, and Kentsfield came out in 2006. Unsurprising given that the Stars architecture that LLano is based on is almost as old. LLano packs a good iGPU though, and if extremely budget-restricted it can be a good choice.

Trinity probably isn't any faster per clock than Llano on average, because remember, it's a 2-module Bulldozer refresh, and Bulldozer was no faster per clock than Stars. We can infer then that Trinity will be of similar performance per clock to Intel's 2006 desktop quad core - the advantage being that you can overclock it to much higher clocks. But honestly, does a 4.5ghz Q6600 really sound all that attractive? It's not going to compete with i5's, because Intel has improved performance per clock 50-100% since 2006.

It will be a good entry level part if priced right, but I feel bad because the part it competes against (Ivy Bridge i3) has a die size less than half as large. It wouldn't surprise me if AMD was actually losing money on every chip sold.

 

[frozentundra123456]

Ahh good, more intel trolls. The more of you that show up trolling AMD threads the better, its a good indication of the threat to intel Trinity really poses.

Someone has to refute the incessant promoting of and by the fanbois on this forum. I

 

[Roland00Address]

But honestly, does a 4.5ghz Q6600 really sound all that attractive? It's not going to compete with i5's, because Intel has improved performance per clock 50-100% since 2006.

Yes a 4.5 ghz Q6600 does sound attractive (assuming the performance scales lineally with clock speed and there is no bottleneck). That is a 87.5% overclock. You are over exaggerating the gains from an i5 2500k vs a q6600

Here are some benchmarks from intel q6600 vs intel i5 2500k (both at stock clocks, info taken from ananadtech cpu bench)

Adobe Photoshop CS4 - Retouch Artists Speed Test Time in Seconds - Lower is Better
25.6 seconds Q6600, 12.6 seconds i5 2500k (i5 2500k is 103% faster)

Cinebench R10 - Single Threaded Benchmark Score in CBMarks - Higher is Better
2778 Q6600, 5860 i5 2500k (i5 2500k is 110% faster)

Cinebench R10 - Multi-Threaded Benchmark Score in CBMarks - Higher is Better
9681 Q6600, 20381 i5 2500k (i5 2500k is 110% faster)

WinRAR 3.8 Compression - 300MB Archive Time in Seconds - Lower is Better
140.7 seconds Q6600, 70.5 seconds i5 2500k (i5 2500k is 99% faster)

Microsoft Excel 2007 SP1 - Monte Carlo Simulation Time in Seconds - Lower is Better
22.1 seconds Q6600, 15.4 seconds i5 2500k (i5 2500k is 43% faster)

--------

So the q6600 is a 2.4 ghz chip. The 2500k is a 3.3 ghz chip, but if it uses all 4 cores and is turboing the max speed for 4 cores is 3.4 ghz (3.5 ghz for 3 cores, 3.6 ghz for 2 cores, 3.7 ghz for 1 core). Thus lets say it has a clock speed advantage of 41% (3.4/2.4=41.6%). If the 2500k is roughly double the speed than it is roughly 41% faster per clock (1.41*1.41=1.99)

 

[Vic Vega]

Wow, "best igpu on the desktop" what a thrill. If you really need a good gpu, add a discrete card. Otherwise almost any igpu is good enough.

Included in the cost of the CPU.

Also, these are targeted at the OEM space for specific form factor. The enthusiast space is looking forward to them, but that's a bonus for AMD.

 

[Yuriman]

So the q6600 is a 2.4 ghz chip. The 2500k is a 3.3 ghz chip, but if it uses all 4 cores and is turboing the max speed for 4 cores is 3.4 ghz (3.5 ghz for 3 cores, 3.6 ghz for 2 cores, 3.7 ghz for 1 core). Thus lets say it has a clock speed advantage of 41% (3.4/2.4=41.6%). If the 2500k is roughly double the speed than it is roughly 41% faster per clock (1.41*1.41=1.99)

41% faster per clock sounds about right, but remember than this 2500K can be taken past 4.5ghz too (more like 4.7-4.8), so it's still at least 41% faster while producing considerably less heat and drawing less power.

I suppose Trinity, if released at ~$100, would have a better priceerformance than a 2500K, and nobody is calling a stock Sandy Bridge slow. Still, that extra $100 buys you 50% more CPU power, and that's per-thread performance too.

 

[Roland00Address]

Here are the numbers for the q6600 vs i7 3770k

Photoshop (same test as above)
25.6 seconds Q6600 vs 10.3 seconds i7 3770k (i7 3770k is 148% faster)
Cinebench (same test)
2778 CBMarks Q6600 vs 6862 CBMarks i7 3770k (i7 3770k is 147% faster)
Cinebench (same test)
9681 CBMarks Q6600 vs 25703 CBMarks i7 3770k (i7 3770k is 165% faster)
WinRAR
140.7 seconds Q6600 vs 57.7 seconds i7 3770k (i7 3770k is 143% faster)
Microsoft Excel
22.1 seconds Q6600 vs 10.0 seconds i7 3770k (i7 3770k is 121% faster)

Now the Q6600 is at 2.4 ghz, the i7 3770k is at 3.5 ghz base but can turbo up to 3.7 ghz for all 4 cores (3.8 ghz for 3 cores, 3.9 ghz for 2 cores, 3. 9 ghz for 2 cores) Thus lets say the i7 3770k has an additional 54% clock speed (3.7/2.4 is 1.54). If the i7 3700k is roughly 145% faster than about 59% is related to the improvements per clock and 54% is related to clock speed. (1.59*1.54=2.45)

 

[Roland00Address]

41% faster per clock sounds about right, but remember than this 2500K can be taken past 4.5ghz too (more like 4.7-4.8), so it's still at least 41% faster while producing considerably less heat and drawing less power.

I suppose Trinity, if released at ~$100, would have a better priceerformance than a 2500K, and nobody is calling a stock Sandy Bridge slow. Still, that extra $100 buys you 50% more CPU power, and that's per-thread performance too.

I rather not take overclock into consideration for that makes things all kinda of complicated. 1) Most people do not overclock 2) Almost all OEMs do not overclock 3) Most trinity chips are going to be sold to OEMs and not enthusiasts 4) trinity can overclock too but to how much and what heat and such, etc. It becomes much simpler if we just talk stock.

I also rather not do speculations on heat and power draw until we get a review instead of speculation. I have no doubts that trinity will use more power/generates more heat than sandy or ivy but that doesn't mean it will sweat heat like bulldozer does. We won't know this fact till we get a review.

Trinity if the price list that is posted on the net is accurate is going be about $110 for the a8 (2 module 4 core) and about $130 for the a10 (2 module 4 core) which puts it in the same territory as the i3s. The i3 3220 tray price is $117, the i3 3225 (hd 4000 graphics) is $134. The i5 tray price is $184 for 3470, $201 for 3475s (hd 4000 graphics), and $225 for 3570k.

 

[Idontcare]

It becomes much simpler if we just talk stock.

I have no doubts that trinity will use more power/generates more heat than sandy or ivy but that doesn't mean it will sweat heat like bulldozer does.

At stock though, bulldozer doesn't "sweat heat", does it?

I was under the impression that bulldozer doesn't gulp the juice until you start overclocking it.

 

[Greenlepricon]

At stock though, bulldozer doesn't "sweat heat", does it?

I was under the impression that bulldozer doesn't gulp the juice until you start overclocking it.

Mine stays extremely cool unless I find a good way to stress it at stock. In the mid 20's to give a rough idea. Even stressed it doesn't go above 40 when water cooled. I use the 8120 at stock 3.1GHz, but have overclocked it to 3.8 without overvolting and even then my temps only raise by around 3 or 4 degrees. It does use more power than you might think, but it's nothing to complain about unless the chip is overclocked. I haven't measured the power draw myself so I can't really help there from personal experience.

 

[ShintaiDK]

At stock though, bulldozer doesn't "sweat heat", does it?

I was under the impression that bulldozer doesn't gulp the juice until you start overclocking it.

That depends on the definition:

 

[StrangerGuy]

That depends on the definition:

Didn't you know AT is a paid pro-Intel adsite?

*sarcasm*

 

[Chiropteran]

That depends on the definition:

Posting that same old graph again? Lets use that benchmark to measure performance, since you think it's so relevant.

Oh look at that, FX-8150 is faster than i5-2500k and within margin of error of i7-2600K

 

[Ancalagon44]

Posting that same old graph again? Lets use that benchmark to measure performance, since you think it's so relevant.

Oh look at that, FX-8150 is faster than i5-2500k and within margin of error of i7-2600K

So it can beat an older architecture while consuming nearly double the power and packing in a few hundred million extra transistors. I can see how that is a victory for AMD.

 

[Gikaseixas]

FX chips performance is all over the chart, sometimes right up there with Core i5/i7 boys and sometimes down there fighting Pentiums. I hope they've fixed some of that with this new cpu. IMO @ $189 the FX8150 is not that bad if the end user doesn't mind the extra voltage, heat and all the tweaking that comes with it.

 

[guskline]

FX chips performance is all over the chart, sometimes right up there with Core i5/i7 boys and sometimes down there fighting Pentiums. I hope they've fixed some of that with this new cpu. IMO @ $189 the FX8150 is not that bad if the end user doesn't mind the extra voltage, heat and all the tweaking that comes with it.

Even better when you can snag one at $170

 

[inf64]

I think 8120 is much better deal right now. And if AMD guys are smart enough,they will price new FX8320 slightly above 8120 and on par with 8150. So you will end up with more power efficient chip(Piledriver) which has somewhat higher IPC and can OC slightly better. It can become a best selling FX8xxx model quite easily.

 

[Phynaz]

Ahh good, more intel trolls. The more of you that show up trolling AMD threads the better, its a good indication of the threat to intel Trinity really poses.

Name calling is the best you could do?

 

[NTMBK]

Name calling is the best you could do?

Coming from the guy who thought "kindergarten class" was the height of wit.

 

[piesquared]

Name calling is the best you could do?

If you don't want to be called a troll, don't be one. So stop derailing the thread troll.


As for Trinity, it will be the perfect chip to build a new desktop with. FM2 compatibility for Kaveri is icing on the cake.

 

[Ancalagon44]

If you don't want to be called a troll, don't be one. So stop derailing the thread troll.


As for Trinity, it will be the perfect chip to build a new desktop with. FM2 compatibility for Kaveri is icing on the cake.

How is calling out BD's failures being a troll? It was garbage.