ARM Co-founder speculates on the future

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
TDP is only one of the criteria, and arguably the least important. I think most smartphone chips have a TDP in the range of 500-800mW.

I'm not totally convinced. Maybe mainstream phones that cost $200. But a brand-spanking-new smartphone with all the bells and whistles? I can believe that with just the CPU core. But add the memory controller, graphics, the arbiter, the internal bus, the L2 cache, no its likely higher than that. TDP is one of the main argument against x86.

Modern desktop chips like Core 2 does not reach TDP so its basically peak. Core 2 needed a specialized program to reach TDP levels of 65W.
 
Last edited:

cheesehead

Lifer
Aug 11, 2000
10,079
0
0
The software infrastructure has a long way to go on ARM before can get anywhere near to rivaling WinTel.

ChromeOS / Linux on ARM is not going to put much of a dent in WinTel in the desktop space.

Microsoft writes a cell phone OS for ARM. Stands to reason that if people have the hardware, they'll make a full desktop OS as well.

How does this cycle time mentioned in the article affect Intel's ability to make inroads using Android? (At this time Intel has Wind River Android and Meego as targets for its Moorestown and Medfield platforms)

The neat trick with the ARM CPUs is that you can shove your entire cell phone on a chip. Not just the CPU, memory controller, and video chipset - the 3G transmitter, bluetooth thingy, analog to digital converters, the whole nine yards. I'm not sure how much of this is actually on the mainstream Qualcomm CPUs and how much is discrete, but seeing as how Qualcomm has been making cell phone hardware for decades, I wouldn't be surprised if the answer was "most of it."

I'm not really that familiar with Intel hardware, but I can say with certainty that they're not going to have the same hardware integration as Qualcomm - if only just because Intel is not a cell phone company.

On the subject of "Qualdroid", I doubt it's going to become the de facto standard in the same fashion as Wintel. Strictly speaking, Qualcomm's processors are just modified versions of a referenced design licensed from ARM, and as a result they're functionally very, very similar to those from Samsung, Apple, and TI. Furthermore, Android is designed to be heavily hardware-agnostic.
 
Last edited:

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
The neat trick with the ARM CPUs is that you can shove your entire cell phone on a chip. Not just the CPU, memory controller, and video chipset - the 3G transmitter, bluetooth thingy, analog to digital converters, the whole nine yards. I'm not sure how much of this is actually on the mainstream Qualcomm CPUs and how much is discrete, but seeing as how Qualcomm has been making cell phone hardware for decades, I wouldn't be surprised if the answer was "most of it."

Ok I'm not sure about Qualcomm, but lots of smartphones have more than just one chip. The PMIC is almost always seperate, and the communication chip is usually seperate as well. Qualcomm's new presentation seems to indicate they are going to integrate that, which means current chips have it on another chip. Memory isn't on die, just stacked. They sure have good packaging tech though.
 

P4man

Senior member
Aug 27, 2010
254
0
0
I'm not totally convinced. Maybe mainstream phones that cost $200. But a brand-spanking-new smartphone with all the bells and whistles? I can believe that with just the CPU core. But add the memory controller, graphics, the arbiter, the internal bus, the L2 cache, no its likely higher than that. .

Those chips are all SoCs. The power numbers are for the entire SoC, which includes IMC, cache, graphics and everything else. I believe the first snapdragons ran at 350mW @1GHz, though its unclear if that is typical loaded power, TDP ( =~sustained maximum power) or peak. Im assuming worst case, so its typical loaded power, which would put the maximum sustained at somewhere twice that.

TDP is one of the main argument against x86.
TDP is just used as short hand for power consumption by people who dont understand all the subtleties. Intel already has an atom with 650mW TDP, thats low enough for a smarthphone. But it doesnt mean it will have a similar battery life as an ARM chip.

Modern desktop chips like Core 2 does not reach TDP so its basically peak. Core 2 needed a specialized program to reach TDP levels of 65W
Yes, there is a lot of confusion about TDP. Its a thermal spec first and foremost, not a power consumption spec.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
As for incentive for software developpers. Seriously. 120 million iOS devices sold to date, 300k applications, 7+ billion downloads. This is the healthiest software eco system out there. And Android is catching up with that rapidly.

Do you have any opinions on the role of app power efficiency on iOS vs Android specifically for the expanding Tablet market?

Recently I have seen some criticism that Android Java apps double hardware requirements compared to apps programmed in Objective-C for iOS.

While "doubling hardware requirements" sounds like a big deal my guess is that even if true it would probably go unnoticed by the typical smartphone user (re: smartphones are mostly used for placing calls not running desktop applications).

But what about iOS Tablet users vs Android Tablet users? Is there a chance any power efficiency difference could become more of an issue?
 

P4man

Senior member
Aug 27, 2010
254
0
0
Do you have any opinions on the role of app power efficiency on iOS vs Android specifically for the expanding Tablet market?

Recently I have seen some criticism that Android Java apps double hardware requirements compared to apps programmed in Objective-C for iOS.

Hmm. I suspect that criticism predates android 2.2 when Java apps where run on a JVM without a JIT compiler. The performance (and therefore, power consumption) hit might have been substantial.

With a decent JIT compiler (as implemented since 2.2), I highly doubt the difference is anything like a factor 2x on a regular CPU, like x86, and compared to the effects of different skills of developers, the difference between Java and C for most apps is probably noise and could swing either way.

If you factor in Jazelle/Thumb which is part of the ARM ISA and specifically designed to run Java/JIT or other run time compiled code (perl, python etc) more efficiently, it might well be the other way around, with Java code beating compiled C, but I cant say I really know.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
TDP. Its a thermal spec first and foremost, not a power consumption spec.

Not entirely true. Core i7 chips Turbo until it reaches the TDP barrier. Up until 2003, both Intel and AMD marked every SKUs with specific power consumption numbers, but after that they went with "family" TDP.

Of course, if you want to put in "work done faster equals faster time to idle" argument into the equation, then it becomes a usage model discussion. But I'm pretty sure when people say "battery life" they include that too.

Intel already has an atom with 650mW TDP, thats low enough for a smarthphone. But it doesnt mean it will have a similar battery life as an ARM chip.

You said something about needing lower standby power/average power/idle power. But how does that work into x86 vs ARM talks? Good idle and standby power will have all the functional blocks off. Average power is a combination of the previous and load power usage.

Chipset part on Pineview and Tunnel Creek take 50% or more of its TDP, but on ARM chips take only 30-40% of the total.
 
Last edited:

P4man

Senior member
Aug 27, 2010
254
0
0
Not entirely true. Core i7 chips Turbo until it reaches the TDP barrier. Up until 2003, both Intel and AMD marked every SKUs with specific power consumption numbers, but after that they went with "family" TDP.

TDP=Thermal Design Power. Its real use is defining what the cooling solution has to be capable off, and that depends on other things than just power consumption, like max ambient temps, maximum allowable temp for the core etc. Yes, its a factor in mobile phones, you cant cool 10W on a fanless device that you put in your pocket, but its not intel's biggest hurdle.

You said something about needing lower standby power/average power/idle power. But how does that work into x86 vs ARM talks? Good idle and standby power will have all the functional blocks off. Average power is a combination of the previous and load power usage.

Not that simple. An x86 core has to do a lot of things an ARM core does not, like complex instruction decoding. Thats why the cpu core of an x86 chip, even one as basic as atom (which doesnt even have OoO execution) is a lot a bigger than an ARM core. Those things consume power too, and you cant just turn them off in idle.

Then there are instruction set extensions, stuff like Thumb that can provide massively better efficiencies when running run time compiled apps. x86 has nothing like it, and I guess thats one reason intel prefers Meego running binary code over Android/Symbian typically running RTC code.

[/quote]Chipset part on Pineview and Tunnel Creek take 50% or more of its TDP, but on ARM chips take only 30-40% of the total.[/QUOTE]

Not sure what you are saying here, nor where you are getting your numbers from.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
TDP=Thermal Design Power. Its real use is defining what the cooling solution has to be capable off,

So what you are saying is TDP has NOTHING to do with power use? I'm pretty sure it isn't the primary role, but plays a very important role in power use. With the same die higher thermals mean higher power use. Of course it might not occur all the time, but even on the Core i5 chip I have here I can see where the CPU's Turbo Mode kicks in even during seemingly not demanding usage scenarios. That'll be even more true on a mobile CPU because its slower.

Unless you think average usage scenarios are really only text based web surfing, with significant amount of reading(idle) between pages. Even if that is true, the TDP levels are generally higher at all power states(HFM/LFM/C0-C6) for a CPU with higher max TDP levels.

Not sure what you are saying here, nor where you are getting your numbers from.

Atom N450: 5.5W TDP which includes the CPU, IMC and GPU
Atom N270: 2.5W TDP just the CPU

Atom E680: 3.9W TDP, includes CPU, IMC, GPU, sound processing, some boot components
Atom Z540: 2.2W TDP just the CPU

TI datasheet OMAP3515:
639mA CPU, 439mA GPU

(Just for comparison, the mA numbers are comparable to the Z500 chip for the CPU)

So on the N450 and E680 chips, the non-CPU parts take 1.7-3W, which is almost as much as the CPU core, if not more, while on the ARM chip, those non-CPU parts take less than the CPU which is already at really low levels.

Thats why the cpu core of an x86 chip, even one as basic as atom (which doesnt even have OoO execution) is a lot a bigger than an ARM core

Bobcat, 2-wide OoO, same TSMC 40nm as A9, 4.6mm2 core, 30% larger than A9's high-perf implementation
 

P4man

Senior member
Aug 27, 2010
254
0
0
So what you are saying is TDP has NOTHING to do with power use?

I said exactly what I said, not the above. Or to quote wiki:

The thermal design power (TDP), sometimes called thermal design point, represents the maximum amount of power the cooling system in a computer is required to dissipate.

Of course that is related to power use, but its not the same.

TI datasheet OMAP3515:
639mA CPU, 439mA GPU

(Just for comparison, the mA numbers are comparable to the Z500 chip for the CPU)

Like I said, Atom (moorestown) already has a low enough TDP to be fitted in a phone. But so far, no one seems in hurry to.

So on the N450 and E680 chips, the non-CPU parts take 1.7-3W, which is almost as much as the CPU core, if not more, while on the ARM chip, those non-CPU parts take less than the CPU which is already at really low levels.

Thats because you are looking at desktop/netbook chipsets. No one in their right minds would use those for a phone. Why would you want a PCI-E bus on a phone?

Bobcat, 2-wide OoO, same TSMC 40nm as A9, 4.6mm2 core, 30% larger than A9's high-perf implementation

Keep in mind most of the die space for the ARM core is the Neon and Thumb units. When running SIMD or FP instructions, I dont know if there is a fundamental advantage for ARM over x86 (or neon over SSEx), but running integer code, there certainly is.

Moreover die size alone doesnt tell you everything, if anything. Even if you can shrink the logic, decoding x86 is simply a lot more complex and that complexity translates into work the CPU has to do. Work is energy.

x86 ISA is also register starved, it implements many more in hardware than it exposes in software, and the cpu therefore has to constantly dynamically rename registers, something that really serves no purpose by itself. That doesnt come free. Lastly, ARM doesnt require branching for simple if then else constructions, making its performance much dependent on complex and power hungry branch prediction.

There is reason intel dropped all of the above issues when they designed itanium.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Bobcat, 2-wide OoO, same TSMC 40nm as A9, 4.6mm2 core, 30% larger than A9's high-perf implementation

Performance per watt would probably make for an interesting comparison here as Cortex A9 @ 2Ghz is sub one watt per core and bobcat has been touted as sub one watt capable.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106


A chart from ARM showing the relative differences between the 3-wide dual core Cortex A15 and the 2-wide dual core A9.

Does anyone know why Cortex A9 is listed twice on that chart (at x5 and x7 relative performance)? Surely the 2Ghz version of A9 is more than 40% faster than the 1Ghz version?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
We will see about that. Frankly, im not so sure. If they port windows 7 or 8 to ARM, who would want it? It wouldnt run any existing windows app, or at least not at acceptable speeds if they rely on emulation.

I'm not sure what to say about the majority of x86 apps out there, but what about MS office? Couldn't that be ported to ARM?
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Does anyone know why Cortex A9 is listed twice on that chart (at x5 and x7 relative performance)? Surely the 2Ghz version of A9 is more than 40% faster than the 1Ghz version?

I think that roadmap has the 2011 part at 1.5GHz. That's a smartphone roadmap anyway.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
More speculation on ARM/Windows 8.

http://www.eetimes.com/electronics-news/4209600/Why-Microsoft-Windows-8-will-run-ARM

Why Microsoft Windows 8 will run on ARM
Rick Merritt
10/14/2010 6:15 AM EDT

SAN JOSE, Calif. – I know the bits on Windows 7 haven't even cooled yet, and here I am speculating on the next big thing already. But the fact is there hasn't been a version of Windows with anything truly hot in it for years and ARM support could be the juice Redmond badly needs.

Windows 8 and ARM is a marriage made in heaven. It would open doors for Microsoft to new kinds of low cost, low power products ranging from consumer tablets for Office weenies to Internet café systems in African villages. At this stage, a new market for a franchise like Windows is manna from above.

It's a good deal for ARM, too. The company has already been pushing beyond the smartphone with its quad-core ready A15 Eagle design that will be available in chips about the time Windows 8 ships.

Recently ARM trumpeted a desktop-like SoC from China. The startup's processor uses a 1.6 GHz dual-core Cortex A9, Mali graphics block and 64-bit memory bus and supports PCI Express, USB and serial ATA. All that's missing is Windows.

It would not be hard for Microsoft to run Windows on ARM. It has runs its Windows CE on ARM for years, so it has intimate knowledge of the hardware.

Microsoft licensed the ARM core recently, but didn't say why. One reason could be it wanted to get intimate knowledge of what a 64-bit ARM implementation would be like for Windows 8.

With ARM pushing toward desktop and even server markets, it will need to deliver a 64-bit core. Again, the Windows 8 timeframe would be about right.

I asked Jeff Chu, director of mobile computing for ARM what he thought about Windows 8 on ARM. "It would be a fantastic thing to see," he said.

Of course Windows on ARM doesn't erase the advantages of the x86, ARM's archrival. Many applications and tools would still have hooks into the x86 that would give Intel (and AMD and Via) an advantage in some markets.

But Windows 8 on ARM is the next major step to leveling the playing field between ARM and Intel in a market where they are bound for head-to-head competition.

So, I fully expect this software will emerge from the oven in about three years. But who knows, by that time a new generation addicted to iOS and Android gadgets might just ask, "What's Windows?"
 
Last edited:

P4man

Senior member
Aug 27, 2010
254
0
0
I'm not sure what to say about the majority of x86 apps out there, but what about MS office? Couldn't that be ported to ARM?

Anything can be ported to ARM, but that hardly guarantees success. Windows was once ported to Alpha and MIPS as well (not too mention Itanium), but they all flopped due to a combination of high prices and poor software availability.

Moving the windows user base to ARM might be more feasible today, but hardly trivial. If people have to give up their wintel compatibility, they may as well chose the best OS, and I doubt for many that will turn out to be windows.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
I've read that article. The article is just an opinion piece at EETimes, not something that's confirmed in any way.

I'm not sure why they'd bother with that. Without the programs, Windows is no different than any other OS. It'd be taking far less resource and time to port that same app to existing OSes like WP7. Putting full Windows onto a ARM platform would cause it to lose lot of power advantages as well, not to mention its far more demanding than Linux-based OSes.

It might make more sense for MS to put their own design on next Xbox for example. Maybe they'll create a portable one? Even their own smartphone CPU?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Here is another article on Windows 8 and virtualization (with more detail).

All this stuff is over my head, but t sounds like MS would like to gut Windows 7 and rebuild into something different.

P.S. I am completely unfamiliar with virtualization, but I have also seen this same concept of "Sandbox" mentioned for Chrome OS.

http://www.tested.com/news/should-microsoft-make-windows-8-a-fully-virtualized-os/581/

Early on in the Windows 7 development process, we learned of something called MinWin. This was to be the most radical change to the Windows codebase yet, trading the aging NT kernel for an almost Linux-style system of compartmentalization. MinWin would represent only the base component of the OS, thus reducing its size, with other components layered on-top. Or so we thought. As development continued, MinWin's purpose shifted, and we learned it was not the compartmentalized wonder we once thought.

But according to some, Microsoft might just be revisiting their compartmentalized vision in an entirely new and radical way. While MinWin was all about splitting up the kernel into individual parts, a recent article from PC Magazine suggests that Microsoft should go a step further - splitting not just the kernel, but the operating system itself, with applications and services running in isolation from one another via virtualization. It sounds crazy, but it might just be what Microsoft needs to reinvent the modern OS.

The ability to simply suspend or start an application could be attractive.

What author Lance Ulanoff suggests is that Microsoft could possibly be turning Windows 8 into one giant virtual machine. Applications would no longer run as we currently know it, side by side in memory with other apps, but inside their own, dedicated virtual machine - a speedy, thin client that is segmented from the rest of the system, but still fully capable of behaving as a normal application. This whole process would, of course, be transparent. You wouldn't actually know you were running numerous, small virtual machines, but that's how the system would be set up.

In essence, it could spawn the ultimate crash-proof OS. Just as you can stop, start and freeze the contents of a virtual machine in VMWare or Parallels, so too could you with a Windows 8 application. This is also beneficial from a security standpoint; there would still be direct hardware access to things like graphics, or external drives, but the sandboxed nature of the machine prevents the rest of the system from being touched. If you're infected by a virus or other unwanted app, "that nastyware simply won't get further than the browser sandbox," explains Ulanoff. Sweet.

Parallels, for example, is a machine that virtualizes all aspects of a system, from BIOS to graphics, overhead that a Windows 8 implementation would theoretically reduce.

Considering Microsoft's previous work on the MinWin kernel, attempts to compartmentalize elements of the Windows OS have clearly been on the company's radar for some time (It's even used to some extent in Windows 7). In fact, if you examine trends in the business market, there might even be a corporate rationale as well. Many servers and data centers rely heavily upon virtual machines these days to split up processing power and hardware access to multiple clients. A perfect example of this is shared web hosting. However, these implementations most often rely on so-called "fat" clients - virtual machines that emulate an entire system, including the hardware and software. This takes a great deal of power, and a lot of fine tuning to keep multiple machines running smoothly.

Companies like VMWare, however, have been working on thin clients in an attempt to solve these problems. Thin virtual machines only virtualize specific components of a system — like the registry or other system variables — while sharing certain hardware components with the host OS. Virtualizing a web server, for example, wouldn't require unnecessary things like graphics or USB devices to be virtualized, reducing the overhead required for the environment to run. That equals more processing power to play with, and more efficient virtualized machines.

Windows 7's XP mode, where old and new coexist, is how Windows 8 should work by default, allowing Microsoft to make radical changes while continuing to provide legacy support.

Even if Microsoft fails to make such a large, sweeping change to their OS strategy, virtualization could still play a large part in other areas. Much as Apple did with OS X and its Classic mode many years ago, Microsoft could do away with the registry and other aging NT components, while still supporting older software through a virtualized layer. It could work very similarly to Windows 7's XP mode, but on steroids, completely seamless and segregated from the core OS.

Of course, every Windows development cycle is the same. Things like MinWin and WinFS, once slated to become core parts of the OS, were eventually dropped or modified in such a way that they were no longer what we expected. And while there may be signs now of a virtualized future, that too is subject to change. One thing is clear, there are big things coming for Windows 8 — it's the only way Microsoft can continue to survive.
 
Last edited:

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Oh boy, looks like another Vista.

Radical modifications seem fine and dandy until they try it and usually fail because they realize that lots of the things are hard to implement and get delayed, which gets features neutered. That's one reason they failed with Vista yet was successful with Windows 7.

Maybe not, but every other new OS MS screws up.
 

P4man

Senior member
Aug 27, 2010
254
0
0
I actually like where windows is headed (if true, and if they pull it off). Its also not as radical as you seem to think; Windows 2008 server already runs on top of hyper-v by default, so its already virtualized.

Then again, nothing about this bares any relation to an ARM port of windows, AFAICT. The only thing that is perhaps or not coincidental is ARM suddenly pushing virtualization in its upcoming A15 core, but it was going to need that anyhow for the server market.

Im not saying I dont believe MS will port windows to ARM; in fact I do think they will. I just dont think that will make it very compelling for consumers. It might have some appeal for servers running Exchange, MS SQL or whatever.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Live from Google Chrome OS event

Some cliffs:

-Apparently the web apps can be run offline
-Gmail is not needed for ChromeOS
-HDMI and USB are not working at the moment, but will work later
-The reference Chrome Book will use Intel, but partners will be able to use any hardware they would like for their own Chrome Books.
-100 MB of free 3G allowance every month for 2 years

Google Chrome OS gets detailed
Google partners with Verizon for free 3G data allowance with every Chrome Book
Google unveils CR48, the first Chrome OS laptop

As far as hardware goes, It will be interesting to see how these Chrome OS products compare to running Windows + Chrome.

For example:

Chrome OS + ARM (battery life, browsing performance, offline app performance)
Chrome OS + Atom (battery life, browsing performance, offline app performance)
Chrome OS + Ontario (battery life, browsing performance, offline app performance)

Windows + x86 (either Ontario/atom) + Chrome Browser + (battery life, browsing performance, offline apps)
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
x86 ISA is also register starved, it implements many more in hardware than it exposes in software, and the cpu therefore has to constantly dynamically rename registers, something that really serves no purpose by itself. That doesnt come free. Lastly, ARM doesnt require branching for simple if then else constructions, making its performance much dependent on complex and power hungry branch prediction.

There is reason intel dropped all of the above issues when they designed itanium.

I'm all for bashing x86 when it's deserved, but out of order execution and register renaming go hand-in-hand. ARM has to be doing it too for A9 and A15. Now, the register pressure does have other consequences (for example, more stack accesses, which have to go through the load/store unit) but register renaming is not one of them. x86 also has CMOV (since i686?); ARM does support much fancier predicated execution, but x86 has a couple useful instructions for avoiding branches.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Any opinions on how RAM usage will factor into performance comparisons of Chrome OS vs Windows 7?
 
Last edited:
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |