backtop


Print 47 comment(s) - last by manicfreak.. on Sep 27 at 1:24 AM


  (Source: AMD)
Chipmaker delivers a 100 MHz bump to almost all of its lineup, its third refresh of the year for some models

Today Advanced Micro Devices, Inc., the world's second largest microprocessor maker, unveiled six new CPUs to hold off Intel while it prepares its next-generation architecture.

First, let's get a quick primer on AMD's CPU brand strategy, which can be a bit confusing.  AMD makes two major lines of CPUs -- a high end line dubbed Phenom II, and a budget lined dubbed Athlon II.  Athlon IIs lack L3 cache and typically feature slightly lower clock speeds.  Also, only Phenom II has a subfamily of hexa-core (6-core) CPUs, code-named "Thuban" -- Athlon II is only available in two-, three-, and four-core variants.  Both processors are built on a 45 nm SOI with immersion lithography process.

The Athlon II lineup has seen 3 refreshes this year (January, May, and now September), which have bumped processor speeds on the various lines by 300 MHz.  The Phenom II lineup has seen only two refreshes this year, though it also saw the launch of hexa-core models.

The latest refresh starts with the hexa-core Phenom II X6.  The Phenom II lineup saw the introduction of its first hexa-core offerings (the Phenom II X6 1055T and Phenom II X6 1090T BE) in April.  A third chip, the Phenom II X6 1035T, began making its way to desktops in May, though it still isn't widely available as an individual CPU.  The latest model, the Phenom II X6 1075T falls precisely between the X6 1090T BE and the X6 1055T in performance, though its a bit more competitive in turbo clock than the Phenom II X6 1090T BE.

Desktop Processors

Core Clock

Max Turbo

Cores

L3 Cache

TDP

Pricing

Phenom II X6 1035T

2.6GHz

3.10GHz

6

6MB

125W

~$175*

Phenom II X6 1055T

2.80GHz

3.30GHz

6

6MB

125W

~$200

Phenom II X6 1075T

3.00GHz

3.50GHz

6

6MB

125W

$245

Phenom II X6 1090T BE

3.20GHz

3.60GHz

6

6MB

125W

~$310















*Not widely available for individual retail.

Next up are Phenom II Quad- (X4) and Dual-Core (X2) designs.  With the introduction of refreshed "Black Edition" models with unlocked CPU multipliers, clock speeds have now been increased 300 MHz since April 2009, and 200 MHz since June 2009.  As with prior launches, the previous top-end models receive a price cut, and the new model jumps in at a similar price point.  These products are seen below.

Desktop Processors

Core Clock

Release Month

Cores

L3 Cache

TDP

Pricing

Phenom II X4 955

3.20GHz

April 2009

4

6MB

125W

~$153*

Phenom II X4 965

3.40GHz

Aug. 2009

4

6MB

125W

~$165

Phenom II X4 970 "Black Edition"

3.50GHz

Sept. 2010

4

6MB

125W

$185

Phenom II X2 550

3.10GHz

June 2009

2

6MB

80W

~$85

Phenom II X2 555

3.20GHz

Jan. 2010

2

6MB

80W

~$90

Phenom II X2 560 "Black Edition"

3.30GHz

Sept. 2010

2

6MB

80W

$105






















The Phenom II lineup generally competes against Intel's new Clarkdale CPUs and its higher end Nehalem models.

Wrapping up, AMD is offering one new higher clocked core to each of its three Athlon II budget lines -- the dual-core X2, the tri-core X3, and the quad-core X4.  These models have seen a roughly 300 MHz clock bump since last September, with three new releases.  The new models bump the previous ones down in price.

Desktop Processors

Core Clock

Release Month

Cores

L2 Cache

TDP

Pricing

Athlon II X4 630

2.80GHz

Sep. 2009

4

4x512KB

95W

~$95*

Athlon II X4 635

2.90GHz

Jan. 2010

4

4x512KB

95W

~$100*

Athlon II X4 640

3.00GHz

May 2010

4

4x512KB

95W

~$105

Athlon II X4 645

3.10GHz

Sept. 2010

4

4x512KB

95W

$122

Athlon II X3 435

2.90GHz

Oct 2009

3

3x512KB

95W

~$70*

Athlon II X3 440

3.00GHz

Jan. 2010

3

3x512KB

95W

~$74*

Athlon II X3 445

3.10GHz

May 2010

3

3x512KB

95W

~$78

Athlon II X3 450

3.20GHz

Sept. 2010

3

3x512KB

95W

$87

Athlon II X2 250

3.00GHz

July 2009

2

2x1MB

65W

~$63

Athlon II X2 255

3.10GHz

Jan. 2010

2

2x1MB

65W

~$64

Athlon II X2 260

3.20GHz

May 2010

2

2x1MB

65W

~$68

Athlon II X2 265

3.30GHz

Sept. 2010

2

2x1MB

65W

$76

































*Not widely available for individual retail.

AMD's Athlon's generally compete with Intel's Nehalem (45 nm, new architecture) and Penryn (45 nm die-shrink) series.

In all this looks like another minor, but solid refresh for AMD as it prepares its mobile low-power Bobcat, higher power desktop/server Bulldozer-based cores, and its new Fusion CPU/GPU die combos.  These designs will compete with Intel's Sandy Bridge CPUs.  Intel may have the edge on performance, but AMD is staying competitive with pricing and aggressive refreshes. 

Of course there's a downside to every upside, and in the case of AMD's refreshes, the downside for consumers is a more cluttered, confusing market.   On the other hand, that seems a fair price to pay for improved performance.

(Price estimates for current models were mostly compiled using Newegg.com's prices.)


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

When will 4 Ghz CPU becomes mainstream...
By nangryo on 9/21/2010 10:46:13 AM , Rating: 2
Not that it needed right now, but, it seems the clock speed has gone forth and back on 1.5ghz to 3ghz clock range for each new architecture. I know that there is still quite lot of improvement from the new architecture alone, not to mention thermal control and power consumption, but why not bump it more with clock speed increase. Is it because of technical limitation or just save later for marketing political reason.




By StevoLincolnite on 9/21/2010 11:04:28 AM , Rating: 1
Some motherboards have a limitation on how much energy they can supply to a processor.
When you ramp up in clockspeed... You generally need to increase voltage depending on the yields and how good the chip is able to clock, which adds to the thermals and power consumption.

As a manufacturing process matures, they can do some tweaks in order to improve clock speeds and other things, which is how we ended up with the Hexa cores that don't want more energy than the previous 965 chips.


RE: When will 4 Ghz CPU becomes mainstream...
By nangryo on 9/21/2010 11:13:39 AM , Rating: 2
Can you give more detail about how much theoretically mainstream mobo could support with current technology? I mean, in overclocked situation, many has gone the 4ghz barrier with not much hassle. Why can't Intel or AMD offer 4GHZ++ processor in their high end tier, that's what makes me curious.


By Motoman on 9/21/2010 11:20:52 AM , Rating: 5
Because they can't get enough yield at those speeds that run through their QA testing flawlessly.

The manufacturers won't release a CPU at a stock clock speed that isn't 100% perfect in every way (granted, Intel has had a couple high-profile math snafus...). When enthusiasts overclock their CPUs, they might encounter minor issues - but generally, they're vastly more tolerant than the manufacturers themselves.


RE: When will 4 Ghz CPU becomes mainstream...
By MrTeal on 9/21/2010 11:40:13 AM , Rating: 4
For a chip, the power requirements scale linearly with frequency, so a 10% increase in freq will give a 10% increase in power consumed, all else being equal. The problem is that all other things aren't equal. For a manufacturing process (45nm, 32nm, etc) there is a certain amount of capacitance in the minimum sized transistor, and when you're switching it you either have to charge or discharge that capacitance. If you want to switch faster, you often have to increase the voltage to be able to move that charge in and out while still meeting your timing requirements. That's what kills you; not only are you switching more often, but it costs you more each time you switch.

Obviously, it is possible to get these chips over 4GHz with extreme cooling and increased voltage. Technically there's no reason why they couldn't bin chips and sell "4GHZ+ capable" extreme editions. There's no reason for them to though, since they can just sell lower clocked EE chips to fill that market segment and then not have to worry about actually supporting chips that will have lower motherboard compatibility and increased failure rates.


By nangryo on 9/21/2010 11:55:28 AM , Rating: 2
Thanks for the detail


By YashBudini on 9/22/2010 10:27:32 PM , Rating: 2
quote:
For a chip, the power requirements scale linearly with frequency, so a 10% increase in freq will give a 10% increase in power consumed, all else being equal. The problem is that all other things aren't equal.

Right, if you look at the same family of AMD processors often you see a voltage increase with higher speeds.

However.

A lot of early releases about a year ago spec'd at 1.4 volts probably with the intention of making them great overclockers at stock voltages. Later processors with higher speeds ended up having the same or even slightly lower voltages. And if your mobo allows it AMDs tend to undervolt rather well, probably for reasons above.


By tjr508 on 9/22/2010 11:50:16 PM , Rating: 3
Frequency will never scale linearly with power. Theoretic gate speed in an FET is proportional to V^(1/2). Power is roughly proportional to V^2. The relationship is a steep quadratic.


RE: When will 4 Ghz CPU becomes mainstream...
By fteoath64 on 9/22/2010 2:31:56 AM , Rating: 2
The over-clocking customer is less than 10% of the target consumer market, so it does not make that much difference in erosion of their margins. Releasing any new product (even if it is a rehash of something old) makes some difference in the sales of the products. At least, it shows that AMD is at it in a big way as Intel has.

Good show AMD, any price drop of existing product line is a good thing. Will drop in on those very soon as the altered prices trickled down the market place. The next gen is a good 6-9 months away from mainstream, so this is a great way to move the market somewhat, especially coming close to the Xmas season.
Coupled with new releases of the 6XXX graphics line, it is a good way to buy AMD products.


By YashBudini on 9/22/2010 10:38:39 PM , Rating: 2
quote:
Good show AMD, any price drop of existing product line is a good thing.

Their price drops are all forced upon them. Good for you, but not so hot from their POV.

quote:
The over-clocking customer is less than 10% of the target consumer market

Maybe so but that's the very target of the BE processors.


By YashBudini on 9/22/2010 10:51:04 PM , Rating: 2
quote:
Can you give more detail about how much theoretically mainstream mobo could support with current technology? I mean, in overclocked situation, many has gone the 4ghz barrier with not much hassle. Why can't Intel or AMD offer 4GHZ++ processor in their high end tier, that's what makes me curious.

It depends on how much power the board needs to supply to the CPU. Many an older AMD board burned up at stock speeds when 140W processors were inserted. Those "limited" boards were designed before the 140W processors came out.

Boards that can handle a lot of OCing are the boards with far more capacitors and voltage regulators. The power needs to be there and it needs to be rock steady and the board handles it, there is no 1.3 or 1.4 volt line coming out of the power supply.

Most mainstream boards are not built like that and will not get to 4 GHz. High OC boards can easily cost twice as much as plain vanilla boards. When the mainstream/stock processor speeds reach 4GHz then just about all boards will support them.


RE: When will 4 Ghz CPU becomes mainstream...
By omnicronx on 9/21/2010 4:45:47 PM , Rating: 2
quote:
Some motherboards have a limitation on how much energy they can supply to a processor. When you ramp up in clockspeed... You generally need to increase voltage depending on the yields and how good the chip is able to clock, which adds to the thermals and power consumption.
That really does not have anything to do with it, chips today actually use less power than chips of 5 years ago, mainly due to smaller processes.

A 6 core Phenom II X6 1090T BE for example only has a vcore of 1.325V, while for example the original A64 was 1.4 or 1.5 vcore.

The fact remains clockspeed is not everything, there are far too many other variables when it comes to system performance. (shorter/more pipelines, cpu cache, hard drive speed, memory, system design in general, etc etc)

There is also the fact that most cpu designs don't scale very well at higher clockspeeds in the first place. The fact that the speed increase ratios decrease as you go higher does not help either i.e a bump for 2ghz to 3ghz (or 33% clockspeed bump will probably be far more noticeable than a 3ghz to 4ghz bump (25%). Of course this does not take bottlenecks at higher frequencies into account either.


RE: When will 4 Ghz CPU becomes mainstream...
By bug77 on 9/22/2010 6:50:10 AM , Rating: 2
quote:
That really does not have anything to do with it, chips today actually use less power than chips of 5 years ago, mainly due to smaller processes. A 6 core Phenom II X6 1090T BE for example only has a vcore of 1.325V, while for example the original A64 was 1.4 or 1.5 vcore.


Cute, but wrong.

The original A64 had a 89W TDP, while the 1090T is a 125W part.

Look here: http://en.wikipedia.org/wiki/List_of_CPU_power_dis...

Btw, vcore is not power.


RE: When will 4 Ghz CPU becomes mainstream...
By Taft12 on 9/22/2010 11:30:30 AM , Rating: 2
When you patronizingly call someone's post "cute, but wrong" you should make sure you understand the topic yourself.

TDP is about heat, not power or voltage.

Like the grandparent was saying, current CPUs really do consume less power even at the same TDP rating due to both smaller manufacturing process and the refinement of design that lowers voltage and clockspeed when a CPU is idle.


RE: When will 4 Ghz CPU becomes mainstream...
By bug77 on 9/22/2010 12:03:30 PM , Rating: 2
Yeah, yeah, I'm sure the P in TDP stands for Heat.


RE: When will 4 Ghz CPU becomes mainstream...
By YashBudini on 9/22/2010 10:29:03 PM , Rating: 3
While they made me laugh the T may have something to do with it.


RE: When will 4 Ghz CPU becomes mainstream...
By bug77 on 9/23/2010 4:17:35 AM , Rating: 2
I feel we're splitting hairs here. Yes, TDP refers to the power dissipated as heat. But think about it: does it make sense for a CPU to actually use less power, while wasting more as heat? If TDP goes up, so does the power the CPU uses.


RE: When will 4 Ghz CPU becomes mainstream...
By MikeMurphy on 9/23/2010 9:46:05 AM , Rating: 2
Dude, TDP and power consumption are only casually correlated. One is heat output and one is power consumption. These are very different metrics and don't ever think you can make assumptions against two different chips sporting different revisions and manufacturing processes.

Furthermore the TDP is more marketing than science and should only be relied upon as a rough ballpark estimate. Also, don't compare AMD and Intel TDPs as they measure them differently.

I need a beer.


RE: When will 4 Ghz CPU becomes mainstream...
By bug77 on 9/23/2010 11:25:26 AM , Rating: 2
Actually, I'm right.

Look here: orph euscom puting. com/ dow nlo ads2/SDA3000AIO2BX-specs. pdf (remove spaces)

Voltage for a Sempron: 1.4V. Amperage: 40.6A. TDP: 59W.
And guess what? 1.4x40.6 = 56.84.

Now for a newer CPU: supp ort.amd.com/us/Pro cessor_TechDocs/43375.pdf

Voltage for a Phenom: 1.2-1.4V. Amperage: 110A. TDP: 125 or 140W.
1.2x110=132. 1.4x110=154.

Therefore:
1. My initial argument is right: newer CPU draw more power than older ones.
2. Your assumption that TDP is only casually correlated to TDP is wrong.


By manicfreak on 9/27/2010 1:24:17 AM , Rating: 2
Wow, comparing the power usage of a single core to a quad-core/hexa-core and you think you got it right? You can't see that one single core is using less power from each successive architecture?


RE: When will 4 Ghz CPU becomes mainstream...
By Aikouka on 9/21/2010 11:20:56 AM , Rating: 2
I'm certainly no expert in the area, but if you look back at the days of the Pentium 4 (Prescott), you'll see that the "megahertz race" was still in full effect. If I remember correctly, it didn't take Intel long to realize that they were hitting a wall by scaling their processors so high. This is when we saw the huge push from single-core to multi-core processors.

I tend to see a lot of overclocks (typically performed on air cooling) of current CPUs max out around 4Ghz (sometimes higher, sometimes lower).

Overall, I'd say that it's probably not in the manufacturer's best interests to attempt to raise their CPUs to their thermal thresholds. Make architectures more efficient or add more processing units to attempt to raise the overall speed, but we know that the latter doesn't always help given the lack of common multi-threaded tasks.


By Mitch101 on 9/21/2010 3:01:37 PM , Rating: 4
I recall reading a long time ago someone at Intel stating the Megahertz race would begin at 32nm. Seems like some truth to that statement.

Intel Sandy Bridge "K" overclocked to 4.9 GHz on air
http://vr-zone.com/articles/intel-sandy-bridge-k-o...


By fteoath64 on 9/22/2010 2:37:09 AM , Rating: 2
Also, to make significant efficiencies in the execution pipeline (ie shorted the pipeline, making more issues per clock), like the micro-Ops fusion improvements in Nahelem. AMD needs to improve these significantly in order to catch-ip and exceed Intel in this area. It seems they have mastered high frequencies but not high execution efficiencies.

ALso, in areas of low-power, there is much to be done. They I say to AMD, why not increase the number of memory channels ?. Like quad channels as opposed to current twin channels ?.


RE: When will 4 Ghz CPU becomes mainstream...
By jrb531 on 9/22/2010 10:38:54 AM , Rating: 2
The Intel P3 was actually faster than the P4 in many ways. Intel decided that the general public was too stupid to realize that MHZ is not the only factor in a CPU and perhaps they were correct.

So they killed off the P3 early to move toward the P4. The P3 did more instructions per clock cycle than the P4 and while I'm not going to say that the P3 had not hit the wall as far as clock speeds... they could have die shrunk it and/or redesigned it to allow faster speeds.

When they did the P4 it was designed for very high MHZ speeds so Intel could proclaim that their CPU's were faster (MHZ wise) than AMD. Rem the entire + ratings AMD added to their chips to "help" educate the public that MHZ is not the sole factor in CPU speed.

IMHO all CPU's should have two ratings... the MHZ and Instructions per clock cycle. You really need both numbers to compare and chip.

This is one of the reasons Intel's CPU's are now faster than AMD even at slower clocks.

If it was up to me I would rate CPU's with two numbers...

1. Single core speed
2. Multi-core speed

A 6 core CPU is wonderful when running applications that can use all those cores but often the tradeoff is a slower "overall" cpu so it would serve the general public to know that the "fastest" speed does not mean a hill of beans for many current apps. Even apps that can use more than one core seldom use all of them or even if they do they seldom, if ever, "fully" use the other cores.

So in this case so many people "choose poorly" when they pick a 6 core 1800mhz cpu over a 4 core 2000mhz CPU - for example. They see the advertized speed or benchmark for some test that "real world" apps NEVER will get even close to and then pick a cpu based on this.

Those "in the know" also realize that the more cores you add, the harder it is to OC because all the cores have to pass at that speed. If you have 6 cores and 5 will do 2400mhz and 1 will only do 2200mhz.... guess what... you're CPU is limited to 2200mhz.

Apps are getting smarter and using cores better each day but nothing and I mean nothing will beat pure speed (adding in instructions per clock of course) and while I agree that you do not want to pass the limit of stability... I'll take extra MHZ over another core anyday - as long as you have at least 2 cores.

What good does extra cores do if they are either not being used or only lightly used? With my Logitech G15 keyboard I can see all 4 of my cores on the display and let me tell you... that 4th core is a "lonely" core... kind of like the Maytag repairman LOL - it never gets used.

The 3rd core seldom gets used either. Sure I bet a few of you can point of App #1 or App #2 that uses all the cores but for the VAST majority of current software... two cores (at the highest MHZ speed) is the best way to go - especially if you are on a budget.


RE: When will 4 Ghz CPU becomes mainstream...
By bug77 on 9/22/2010 11:14:00 AM , Rating: 3
quote:
The P3 did more instructions per clock cycle than the P4 and while I'm not going to say that the P3 had not hit the wall as far as clock speeds... they could have die shrunk it and/or redesigned it to allow faster speeds.


They did. It was called Banias, then Dothan, then Conroe.


By mino on 9/22/2010 5:11:06 PM , Rating: 2
Umm, not really.

Banias was in fact an emergency re-incarnation of PIII AFTER P4 has proven useless for mobile applications.

Until Banias, the only thing stopping AMD from ruling the mobile roost was Intel's bullying of vendors not to use them.

Had it not been for some clever folks at Izrael who started Banias on their own mostly, in 2003 Intel would be DEAD IN THE WATTER.


By Indigo64 on 9/21/2010 6:19:52 PM , Rating: 5
So in this day and age people are still stuck on Megahertz and Gigahertz?

Gone are the days of "stupid engineering" where marketing rules the roost. Computer technology of today is all about "smart engineering" and making things work within a predefined power and thermal envelope, not to mention architectural improvements that can increase processor performance gains in ways pure clockspeed could never do.

Look at the Atom. Would anyone have bought an in-order execution core 5 years ago? The K6-2 was an in-order engine and look how much it was panned back in its day.

Also look at how fast "mainstream" chips are - 2.2 up to 2.8 GHz, hell there are people who have 2.66 GHz i7's that are very happy with them. AMD users have their 2.6's and 2.8s as well.

I'm very happy with where the market is. Clockspeed doesn't mean everything. People who are still stuck on that tired old mantra will be very disappointed when their CPUs don't hit 10 GHz in 2 years.


Fact check
By Hyperion1400 on 9/22/2010 6:33:33 AM , Rating: 2
quote:
Today Advanced Micro Devices, Inc., the world's second largest microprocessor maker


I expect that kind of comment of a website like Gamespot, but a site that calls itself Dailytech, and typically does a good job of reporting tech news(not so much other stuff :P), shouldn't make such simple and obvious mistakes. IIRC, AMD is the 10th largest microprocessor manufacturer overall, and the second largest x86 microprocessor manufacturer.

I hate to sound like I am riding you Jason, it's just that things like that seem to irk me. I do the same thing when people confuse clips with magazines...




RE: Fact check
By mino on 9/22/2010 5:15:21 PM , Rating: 2
Well, by mainstream definition, they are indeed 2nd.

Most people do not see ARM-based SoCs as microprocessors.


RE: Fact check
By YashBudini on 9/22/2010 10:31:57 PM , Rating: 2
When was the last time anything with an ARM advertised its capabilities? No wonder they don't know.


A very nice refresh, but...
By quiksilvr on 9/21/10, Rating: -1
RE: A very nice refresh, but...
By jrb531 on 9/21/2010 11:04:52 AM , Rating: 5
Tell ya what... I'm on my 4th CPU for the same motherboard with AMD. Sure Intel is better but in the age of the Video Card (NOT CPU!) how many FPS does an INtel CPU get over and AMD?

Intel makes great CPU's (this round) and AMD makes good CPU's but presently Intel is better but I'll tell you what saves me money now...

AMD's are cheaper, fast enough, and allow me to pop in a new CPU on my old 2-3+ year old motherboard without changing anything else.

I can keep my 8gig of fast DDR2 and my rock stable motherboard and not have to rebuild my computer just to change in a new CPU.

How long does Intel keep a socket alive before then change it? Intel "could" keep the sockets the same and if they had to upgrade them allow their chips to work in older motherboard with slightly less abilities (like AMD did whith their AM3 CPU's working in older AM2+ boards) but they don't want you swapping out just the CPU's when they want you to buy a new motherboard with their own MB Intel Chipset.

Look at what Intel does for their laptops... while I would prefer an Intel CPU in my laptop... not at the price of having to run that crappy Intel built in video! I'll take a slower AMD CPU with built in 4200 Video anyday over the better Intel CPU with that horrid Intel video.


RE: A very nice refresh, but...
By Reclaimer77 on 9/21/10, Rating: 0
By StevoLincolnite on 9/21/2010 2:00:20 PM , Rating: 4
quote:
Going from the Core2 to the i7 adds something like thousands of more pins to the CPU


They did?
Socket 775 has 775 "pins" (They aren't pins anymore, more like pads).
Socket 1156 has 1156 pins.
Socket 1366 has 1366 pins.

So in a worst case scenario Intel added 591 "pins" since the Pentium D/Core 2 Duo.
Best Case (a-la 775 vs 1156) is a difference of 381 pins.

That is hardly "thousands".

To be honest, I don't expect the latest Intel processors to be compatible with 775, with the entire moving of the North Bridge on-die and such.
But you must admit Intel does seem to enjoy changing sockets around to almost force you to spend more money to get a new motherboard with it's own chipset.


RE: A very nice refresh, but...
By StevoLincolnite on 9/21/2010 11:05:50 AM , Rating: 2
At Idle they are competitive with Intel with power consumption.
And lets face it, the majority do not run a processor flat out 24/7.

Plus... There is always undervolting, think of it as the reverse of overclocking.


RE: A very nice refresh, but...
By xti on 9/21/2010 11:25:29 AM , Rating: 2
not to mention it doesn't make sense that someone who is paying a premium for a high end chip where watts start to get on the high end...worrying about light bills down the road.


RE: A very nice refresh, but...
By Reclaimer77 on 9/21/2010 12:35:05 PM , Rating: 1
What I don't understand is why 100Mhz? It's such an absurdly small increase that the difference is entirely inconsequential. When it comes to it's impact in performance, 100Mhz in any modern CPU simply does nothing for you.

As far as power consumption goes, I honestly doubt a 100Mhz OC would register a measurable amount in increased wattage. You can easily get that out of a well designed chip like these without even increasing voltage.


RE: A very nice refresh, but...
By twhittet on 9/21/2010 1:39:06 PM , Rating: 2
What I understand is - why not? 100Mhz is 100Mhz. It's a round number. It's the third time they've done it this year, plus prices are either stable or even cheaper. It's a tight market - I'm glad AMD is attempting to be competitive.


RE: A very nice refresh, but...
By Reclaimer77 on 9/21/10, Rating: -1
RE: A very nice refresh, but...
By Mojo the Monkey on 9/21/2010 4:53:24 PM , Rating: 2
What?! Yes, you will see performance gains. Maybe not appreciable gains to the end user if they are literally moving from 2.8 to 2.9 ghz in an otherwise identical chip. But benchmarks will show some change.

And your comment about multiplicative performance makes no sense in relation to an explanation about no gain from 100mhz.

Z x 3100 = Z x 3200??? I think not... unless Z=0. Well there you have it. You will see zero performance gains when your computer is idle or switched off. Bravo.


RE: A very nice refresh, but...
By Reclaimer77 on 9/21/2010 5:42:40 PM , Rating: 2
quote:
Maybe not appreciable gains


Well duh, what did you think I meant? You might see 1 FPS increase in a benchmark, or a 2 second improvement in encoding times or something like that. But 100Mhz? It's barely going to register.

This is a factory overclocked chip. But ask any experienced overclocked if he would bother doing an OC for 100Mhz. You would get laughed at.

Oh and I meant additive, not multiplicative, I put the words in the wrong order. Curse you invisible edit button!


RE: A very nice refresh, but...
By Hyperion1400 on 9/22/2010 6:40:36 AM , Rating: 2
Actually, performance gains through increasing clock speeds are more non-linear, diminishing return than they are additive. IE 100% boost to clock speed =/= 100% performance boost


RE: A very nice refresh, but...
By YashBudini on 9/22/2010 10:35:45 PM , Rating: 2
More so with AMD and their memory frequency divider nonsense. The situation really shows when going from 2 to 2.1 GHz and DDR2-800 memory.


RE: A very nice refresh, but...
By bug77 on 9/22/2010 6:43:08 AM , Rating: 1
Actually, you got it backwards.

Intel cites "typical power usage" whereas AMD cites "maximum power usage". The difference is so significant that I had to replace my power supply when going from an AMD 89W rated CPU to an Intel 65W; it couldn't handle the additional load.


RE: A very nice refresh, but...
By YashBudini on 9/22/2010 11:03:09 PM , Rating: 2
quote:
I had to replace my power supply when going from an AMD 89W rated CPU to an Intel 65W; it couldn't handle the additional load.

I looked up the word marginal and saw a picture of your power supply. No offense.

Informative and comical article about power supplies:
http://www.jonnyguru.com/modules.php?name=NDReview...

Actually it's not comical if you own one of those FAILed power supplies, but it's these power supplies that force video card makers to state ridiculously high power rating requirements for their video cards. There are simply too many PSs that can't deliver.

A 750 watt power suppply for $18? Yeah, sure it is.


RE: A very nice refresh, but...
By YashBudini on 9/22/2010 11:11:21 PM , Rating: 1
quote:
Still takes more power consumption than the competition. I'm sorry, but I'm willing to spend a little extra because I'll get that money back and then some even if the difference is a saving of 10 watts.

If your PC is an email pc it doesn't matter. If your running a heavily loaded server(s) or you're folding then it does matter. For everyone else it's a matter of time whether it's worth it or not. But yes what you stated is correct, despite the rating, I think you just annoyed people who missed these factors is all.

But given that you should also be looking at PS efficiency. And non-APFC PSs should have been outlawed long ago.


RE: A very nice refresh, but...
By YashBudini on 9/22/2010 11:19:20 PM , Rating: 2
quote:
Rating: 1

What? Just me? What about the horse I rode in on?


"I modded down, down, down, and the flames went higher." -- Sven Olsen














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki