Intel keeps up the unethical SDP scam with “new” 4.5W parts [S|A]

[CPricecrispi]

It's called marketing. Point out the selling points and benefits, and make the a shitty parts look good. Intel can change their TDP or SDP definition, warp benchmarks, or manipulate facts. It's part of the marketing business and if they didn't try, they ain't working hard enough to shine that turd.

Intel has nothing in the RISC ARM market except netbooks and those are disappearing like dinosaurs. It's all Samsung/ Apple, Qualcomm Snapdragon for Andriod, and Nvidia in some tablets. They can try to manipulate numbers but once it is in a test unit, Intel's architecture will show it can't hang with the big 2 ARM manufacturers.

 

[AtenRa]

Dude. TDP does not have an industry wide, common definition.

Intel COULD use TDP here with the same figures and guess what, it would be just as meaningless; Even if Intel used TDP they could say whatever the heck they want. The manufacturer determines TDP, you act like it has some sort of industry wide standard. That is not the case. SDP and TDP do the same thing. Determine how versatile a cooling system must be during normal applications of their choosing.

There is a difference between TDP and SDP and that’s the applications they used to measure each at different TJ temps. You cannot use the SDP number for the same applications used to obtain the TDP number. Also, performance will be significantly lower at SDP than TDP.

So, an OEM can use the SDP numbers to create a device, but that device will never get the same performance it would using the TDP numbers. Its like under-clocking your CPU, the device will never get at higher clocks because it will throttle down constantly when different applications are run in the device than those measured for the SDP.

And since Intel doesn’t say what application they are using to obtain the SDP number, performance cannot be compared against the competition. But, Intel can use the lower power number (SDP) for marketing against the competition. They can use the TDP performance and say the device is using a SDP processor.

 

[NTMBK]

It is obvious. These low end chips are designed to run on low power and so that is how they will perform like all the other ARM tablet junk. If you want a fast processor buy an i-7 or an i-5. Buy the kind of processor designed for what you want to do.

Due to Intel's marketing stupidity, they will use the same i7 brand for their 11.5W parts as they will for their 130W parts.

 

[NTMBK]

Then why is a 4.5w SDP remotely relevant to the cooling solution in place? Oh wait, it isn't.... SDP only dictates battery life perhaps, TDP dictates how large the cooling solution must be. SDP could be 0.1mW, if the TDP is 15W that sucker needs a fan.

Not necessarily. Intel has plenty of thermal throttling technology- just set the right limits, and the chip will throttle to prevent meltdown. Just be prepared to see reduced performance.

 

[NTMBK]

HP seems to want to try. http://www.engadget.com/2013/06/06/hp-fanless-intel-haswell-tablet/

Sounds exciting. A good fanless tablet, which is dockable with a keyboard to make it into a netbook, with a minimum 1080p screen is what I really want from my next laptop. My C60 netbook is a lovely little thing, but I could do with a higher resolution screen and a little more grunt. I was expecting to wait until Broadwell, but if they can pull it off with Haswell I will be impressed.

 

[FwFred]

There is a difference between TDP and SDP and that’s the applications they used to measure each at different TJ temps. You cannot use the SDP number for the same applications used to obtain the TDP number. Also, performance will be significantly lower at SDP than TDP.

So, an OEM can use the SDP numbers to create a device, but that device will never get the same performance it would using the TDP numbers. Its like under-clocking your CPU, the device will never get at higher clocks because it will throttle down constantly when different applications are run in the device than those measured for the SDP.

And since Intel doesn’t say what application they are using to obtain the SDP number, performance cannot be compared against the competition. But, Intel can use the lower power number (SDP) for marketing against the competition. They can use the TDP performance and say the device is using a SDP processor.

What's the problem? Test the OEM device. Same reason I don't look at Anand's Snapdragon 800 review and draw too many conclusions. I'll wait until I see how it performs in a device. The same will apply to any fanless Haswell.

If I was to buy one, it would be because it can operate in fanless, but can dock and get cooled with extra performance. Dynamic range is a nice feature, why is there such an issue with a 4.5W operating point?

 

[sushiwarrior]

Not necessarily. Intel has plenty of thermal throttling technology- just set the right limits, and the chip will throttle to prevent meltdown. Just be prepared to see reduced performance.

But my magical definition for SDP means power consumed while turned off, look at how low it is compared to everyone else! This is a revolutionary breakthrough.

SDP is a meaningless number, it's like saying my 7990 has a "CIDP" of 5W - crossfire idle design power. Meaningless numbers making meaningless claims

 

[NTMBK]

But my magical definition for SDP means power consumed while turned off, look at how low it is compared to everyone else! This is a revolutionary breakthrough.

SDP is a meaningless number, it's like saying my 7990 has a "CIDP" of 5W - crossfire idle design power. Meaningless numbers making meaningless claims

SDP is a meaningless number. I have said this myself multiple times in this thread. Where did I mention SDP in my post?

Thermal throttling is a well established feature, and it will prevent full on meltdown of a device. Saying otherwise is hyperbole. But it will limit performance.

 

[sushiwarrior]

SDP is a meaningless number. I have said this myself multiple times in this thread. Where did I mention SDP in my post?

Thermal throttling is a well established feature, and it will prevent full on meltdown of a device. Saying otherwise is hyperbole. But it will limit performance.

But you're suggesting that thermal throttling is based on an arbitrary SDP limit, which doesn't have a clear definition, and is exceed-able. Without knowing what defines the SDP (50% CPU load? 75%? 25%?), what is the difference between thermal throttling with a 4.5w SDP and thermal throttling with an 11.5w TDP. Both processors exceed their cooling at the same amount of load...

My point is, I don't see there being any difference between a 4.5w SDP processor and one without an SDP. Unless core power states are changed, and even then that will just cripple performance, it doesn't change anything.

 

[NTMBK]

But you're suggesting that thermal throttling is based on an arbitrary SDP limit, which doesn't have a clear definition, and is exceed-able. Without knowing what defines the SDP (50% CPU load? 75%? 25%?), what is the difference between thermal throttling with a 4.5w SDP and thermal throttling with an 11.5w TDP. Both processors exceed their cooling at the same amount of load...

My point is, I don't see there being any difference between a 4.5w SDP processor and one without an SDP. Unless core power states are changed, and even then that will just cripple performance, it doesn't change anything.

No I'm not! For goodness sake, I am not talking about anything to do with SDP! I am talking about thermal throttling, which is based on temperature- a different quantity to heat generated, which is what TDP/SDP are talking about.

 

[sushiwarrior]

No I'm not! For goodness sake, I am not talking about anything to do with SDP! I am talking about thermal throttling, which is based on temperature- a different quantity to heat generated, which is what TDP/SDP are talking about.

Yes, thermal throttling will be the same with any chip, that does not mean that a chip with an SDP of 4.5w will not need a heatsink which can handle more than 11w (else it will experience extreme throttling/thermal overload under stress...)

All this means is that SDP doesn't actually change anything about the thermal requirements/cooling solution used for the chip, IMO.

 

[Dresdenboy]

My point is, I don't see there being any difference between a 4.5w SDP processor and one without an SDP. Unless core power states are changed, and even then that will just cripple performance, it doesn't change anything.

The SDP spec also includes a lower Tjunction than SDP-less CPUs. This implies, that a SDP CPU would throttle at lower temps, thus limit performance.

 

[ShintaiDK]

The SDP spec also includes a lower Tjunction than SDP-less CPUs. This implies, that a SDP CPU would throttle at lower temps, thus limit performance.

It will only limit performance if its not properly cooled.

 

[Arkaign]

With that attitude you should find yourself saying the same of AMD's TDP.

Is this true, that you feel the same about the relevance of AMD's 125W TDP specification for the FX-8350 as you do the 4.5W SDP specification of Intel's future Haswell SKU?

Outstanding question. Your personal hard work in testing was invaluable. I still recommend the 6300, 8320, and 8350 depending on circumstances, but your detailed testing that revealed the actual power profile of that product was incredibly detailed and enlightening.

I am not exaggerating in the least to say that I think you have been the most valuable single member of a forum that I've ever witnessed. I may be a hothead from time to time, but I deeply respect and appreciate the kind of attention to detail that you provide. I also see a direct alignment between intelligence and impartiality. The higher the intelligence of a person, the less likely they are to become drones that allow favoritism to get in the way of pure, unadulterated logic and fact.

 

[Idontcare]

Outstanding question. Your personal hard work in testing was invaluable. I still recommend the 6300, 8320, and 8350 depending on circumstances, but your detailed testing that revealed the actual power profile of that product was incredibly detailed and enlightening.

I am not exaggerating in the least to say that I think you have been the most valuable single member of a forum that I've ever witnessed. I may be a hothead from time to time, but I deeply respect and appreciate the kind of attention to detail that you provide. I also see a direct alignment between intelligence and impartiality. The higher the intelligence of a person, the less likely they are to become drones that allow favoritism to get in the way of pure, unadulterated logic and fact.

Thank you for the extremely kind words :$

But no one is immune from having moments of weakness and getting a little hot-headed though. My alien blood may not be red, but if you cut me I still bleed, and if you cross me on a bad day I will have a strong desire to verbally throttle you for it...as we all put our pants on three tentacles at a time. (all hail gallaxhar )

 

[AtenRa]

Dynamic range is a nice feature, why is there such an issue with a 4.5W operating point?

As I have explained before, my only problem with the SDP is that Intel doesn’t allow to ask questions about it. I want to know at what applications the processor will transition to the SDP point. I want to know all the specs invovled.



Is that too much to ask from Intel ??

 

[FwFred]

As I have explained before, my only problem with the SDP is that Intel doesn’t allow to ask questions about it. I want to know at what applications the processor will transition to the SDP point. I want to know all the specs invovled.

I would like to see some detailed specs as well, but I think due to the impact of different workloads and the dynamic thermal management of Haswell it would be very difficult to spec. Intel cannot hide once these devices show up in fanless tablets, so I think we just need to be patient and wait for one to be reviewed by Anand.

I don't think this is significantly different than mobile SOC vendors. Tegra 4, Snapdragon 800 (especially in a phone), Exynos Octa also require dynamic thermal management and would not run at the rated frequency for all workloads in a constrained thermal environment.

I see SDP as a response to the mobile ecosystem, and tellingly Intel only lists it for the Y series.

 

[tential]

I would like to see some detailed specs as well, but I think due to the impact of different workloads and the dynamic thermal management of Haswell it would be very difficult to spec. Intel cannot hide once these devices show up in fanless tablets, so I think we just need to be patient and wait for one to be reviewed by Anand.

I don't think this is significantly different than mobile SOC vendors. Tegra 4, Snapdragon 800 (especially in a phone), Exynos Octa also require dynamic thermal management and would not run at the rated frequency for all workloads in a constrained thermal environment.

I see SDP as a response to the mobile ecosystem, and tellingly Intel only lists it for the Y series.

Yup this. Intel simply trying to show that Haswell can work in lower power devices, doing the things they normally do.

Either way, as stated before, the numbers are irrelevant. Wait til we get the actual chips. Are we seriously going to throw fits over marketing slides?

When reviewers get the products, and release benchmarks, then we care.

 

[Idontcare]

As I have explained before, my only problem with the SDP is that Intel doesn’t allow to ask questions about it. I want to know at what applications the processor will transition to the SDP point. I want to know all the specs invovled.



Is that too much to ask from Intel ??

Currently that is too much to ask of any MPU manufacturer or OEM reseller. Be it x86, Intel or AMD, or ARM or GPU, etc.

What motivation does Intel have to be any more transparent than AMD or Nvidia or Qualcomm or Apple?

 

[AtenRa]

Currently that is too much to ask of any MPU manufacturer or OEM reseller. Be it x86, Intel or AMD, or ARM or GPU, etc.

What motivation does Intel have to be any more transparent than AMD or Nvidia or Qualcomm or Apple?

Well, if WE (the customers, the press etc) don’t ask and "motivate" all those manufacturers things will get even worst.

 

[CHADBOGA]

Well, if WE (the customers, the press etc) don’t ask and "motivate" all those manufacturers things will get even worst.

Yet amazingly you seek to only "motivate" one manufacturer.

Funny that.

 

[AtenRa]

Yet amazingly you seek to only "motivate" one manufacturer.

Funny that.

I was "Motivating" AMD with their High prices of the HD79xx series, I’m sure you’ll find more “Motivation” somewhere around this forum or others.

 

[Aikouka]

As I have explained before, my only problem with the SDP is that Intel doesn’t allow to ask questions about it. I want to know at what applications the processor will transition to the SDP point. I want to know all the specs invovled.

I guess I still don't see the point in knowing it. What benefit is it going to gain if they spill the beans on the entire SDP qualification process? Like I mentioned earlier, we're probably never going to be implementing these devices directly. At best, we're most likely going to wait for a site like Anandtech to post battery life and performance results for a product that implements one of these low-power Haswell processors. At least in my opinion, those are numbers that we really care about.

 

[pelov]

At best, we're most likely going to wait for a site like Anandtech to post battery life and performance results for a product that implements one of these low-power Haswell processors. At least in my opinion, those are numbers that we really care about.

You know that how, exactly? Because you derived it from an arbitrary SDP number that provides nothing of meaning?

Coincidentally you ended up answering your own question. Apparently you don't see the point in criticizing them for these worthless marketing gimmicks but you're already assuming that the devices will perform better or even sip less power than previous Haswell/Ivy tablets ... and all that derived from a magical SDP number which nobody knows what it means!

But perhaps the press release was meant just for the OEMs :sneaky:

Intel Marketing 1 - Aikouka 0

 

[Keysplayr]

You know that how, exactly? Because you derived it from an arbitrary SDP number that provides nothing of meaning?

Coincidentally you ended up answering your own question. Apparently you don't see the point in criticizing them for these worthless marketing gimmicks but you're already assuming that the devices will perform better or even sip less power than previous Haswell/Ivy tablets ... and all that derived from a magical SDP number which nobody knows what it means!

But perhaps the press release was meant just for the OEMs :sneaky:

Intel Marketing 1 - Aikouka 0

At least its pretty easy to see when someone protests far too much for something so meaningless. Above and beyond, I believe its called.