i7 620M vs. 720QM

Ahh so for everything the cpu does, there are many pieces of executions needed to be done, so HT allows both threads to utilize the resource in that 1 physical core to speed up the process, like multitasking

 

Hi guys, I'm looking at M15x and the choice between the 620m and 720qm. The 620 is an extra £120, thats $177, I don't think its worth that?

 

I checked the U.K. M15x pricing; base CPU is the i3-330M, and
+£90/$133 i5-430M
+£140/$207 i5-520M
+£160/$236 i5-540M
+£180/$266 i7-720QM
+£300/$443 i7-620M
+£590/$871 i7-820QM
+£1070/$1580 i7-920XM

Based on those prices, the last three are very overpriced and I definitely wouldn't bother with those.

You're still paying more than you should for the i7-720QM, but it's less overpriced than most of the other options, and it would be nice if you want the four cores for tasks like video encoding. However, if you're concerned about battery life you're better off sticking with an i5; they'll do the job for gaming anyway.

On the whole, I'd say you should be choosing between the i5-430M, the i3-330M, and the i7-720QM. The rest are poorly priced.

 

Thanks matey, think i'm going to go with the 720qm. So like you said, I can multi task and also easier to upgrade later

 

They all use the same socket, so I don't think there would be a difference in how difficult it would be to upgrade later. In fact, if that's what you want to do, you should spend less now.

As for multitasking, all those CPUs will do a great job. The majority of the time you won't really fully load all of the cores unless you're doing a single task that can use them all; even then, the i7-720QM is only around 20-30% faster than, say, the i5-430M. What kind of usage are you referring to when you say "multi task"?

 

It all comes down to what you plan to use it for as well. If you plan to play games and do every day tasks then it won't really matter. But if you want to play specific game titles or do emulation or rendering then your choice will be different.

Sure cost is important but you should also look at what is most important in terms of your activities.

 

I would go with the i7 720m, it's best to keep up tech. It'll be much more useful later on with quad cores than with dual cores. I like the i7-820qm personally, it's must faster than 720qm although much more expensive =p

 

Read the 620M vs. 720QM and 540M vs. 620M reviews here on NBR (ie. the first posts of this thread) and notebookcheck respectively. There is insignificant difference in performance between the three, even in multithreaded applications. So no, the 720QM will never be more useful than the dual cores. In the end it comes down to price, the cheapest (heck even the 520M) will be sufficient and most cost effective for performance.

 

People have been saying things like that for years, myself included but I don't see any real evidence of that. The only game that seems to care about Quad Cores is GTA IV but even in that case, I have seen people run it perfectly with a 3.0 Ghz Core2Duo.

When BFBC2 was first released, only people who owned Quad Cores could get good performance and I thought that was a sign until they optimised the Code where it played just as well with Dual Core. I currently own the i7820qm but I have an i7 620M on the way so I will see.

 

It's nice to see a rational thought process every now and then Even from a hardware perspective, more cores != better. Remember, we are discussing laptops, which have limited power envelopes and limited thermal capabilities. In order to utilize more cores, you have two non-mutually exclusive options: 1) increase the power envelope or 2) reduce the clock for each core to compensate.

Of course, there is some overlap, but the idea is straightforward: you have to make a compromise.

Therefore, one needs to consider the type of applications they expect to encounter with their system. What is more meaningful, raw clock, or multiple thread capability? This question is not so straightforward. Multi-threaded programming is not always feasible, so there will be times when you cannot exceed the use of a dual or even single core, and a higher clock can be more advantageous (all other factors, such as architecture, maintained as equal). In this case, a CPU with less cores tends to have the advantage, because less cores implies less complexity, reduced losses, and a greater potential for a higher clock within the respective power/thermal envelope.

In multi-threaded scenarios, it really depends on how much of a clock penalty you pay for having more cores. For example, if you can only run a quad core processor at 1.5 GHz with all cores active, in comparison to a dual core processor at 2.8 GHz, the advantage may not prove significant.

Suffice to say, there are many factors to consider, and there will rarely be a clear winner. It all depends on an individual's priorities. Perhaps power savings or cost is more important than a 10% (arbitrary value) performance increase for a limited number of applicable scenarios. Wherever there is a constraint, there must be a compromise, so it is up to the end user to decide where to prioritize the factors that yield the optimum result.

 

Can anyone comment on the 620M v 720M with regards to virtual machine performance?

I do heavy software development and run virtual machines on a daily basis. Currently considering 620M on a Dell Latitude E6400 or stepping up to a 720QM on a Dell Precision M4500. The cost differential, in sum total, is about $600 dollars for the configurations that I'm looking at.

I currently have an E6400 and it was doing pretty good until it had the displeasure of meeting a Mr. SharePoint Server 2010 (still usable now, but definitely sucking air!). I'm not entirely keen on the idea of lugging around a larger laptop, but if it meant significantly better performance for virtual machines, I'd consider it.

Anyone know how VMWare Workstation views the logical cores? For example, you can configure the number of cores to assign to VMWare. Does it use the logical cores or the physical cores when you assign the processor count?

Thanks.

 

I have an E6500, and it sucked a bit of air with the larger VM's as well, until I put an SSD in it. Are you running an SSD or HDD on your 6400?

 

some real results here
http://forum.notebookreview.com/msi...7-total-ghetto-buget-build-attempt-cnd-3.html

 

I had the opportunity of testing out GTA IV with the i7 620M vs the i7 820QM on the same system, the same settings and it would appear that even though the i7 620M was at a consistent 3.06 Ghz Turbo Boost vs the i7 820qm at only 2.0 Ghz per Core, the Dual Core simply could not cope at all.

It had severe stutter, lagg and huge drops in frames even when the settings were lowered. The 3.06 Ghz Dual Core had CPU usage of 80% while the Quad had usage of only 40-50% or lower. The benchmarks between the two showed very similar frame rates but when you played the actual game, only then did you notice the immediate difference.

I also found the i7 820qm to perform the same in Dual Core tasks such as PS2 and Gamecube emulation as the Dual Core did with exactly the same frame rates, even though the i7 820qm was only turbo boosting to 2.66 Ghz vs the constant 3.06 Ghz of the i7 620M.

Of course the i7 720qm might be a totally different story in GTA IV when Vs the i7 620M but I doubt it since those extra Cores are what really matters.

 

Interesting. I'd like to see some more in-depth testing, but the GTA 4 results are quite strange. From what I've heard, it's quite badly behaved due to poor porting from the console version. What I would really like to see is a graph of framerates over time running the exact same in-game scene; it's a pity so few people bother to make graphs. Average framerate is indeed quite lacking as an only metric, and while minimum framerates are nice they're not that reliable. Graphs are so much nicer.

The fact that the i7-820QM only offers ~30% more raw processing power makes the CPU usage figures quite strange as well, considering that 80% usage of the dual-core should offer significantly more computational power than 40-50% of the quad-core.

 

I was convinced that the 3.06 Ghz Dual Core i7 620M would give at least the same level of performance as the i7 820qm with just a small amount of stutter.

With the i7 820qm, I had everything on high, AF at 16x, Vsync disabled and all sliders at 60, latest 1.070 patch and frame rates were from 30-40 fps and sometimes higher with occasional drops down to 25 fps (huge explosions) but not for very long.

With the Dual Core using those same settings, the fps was at a constant 20 - 21 fps with a constant stutter. For example, when you moved the camera around it was like slow motion. Even when I vastly lowered those settings, there was not much difference. Only when I disabled Night shadows did I notice a bigger difference but still not enough.

With the Quad that stutter simply did not exist.

The benchmark results for both CPU's were only a few fps apart which is why I say benchmarks are unreliable in this case. Frames per second simply do not account for how smooth the gameplay is with this game.

You could get 30 fps with or without micro stutter but the benchmark won't show this.

Maybe on Desktop based systems the CPU power of a Dual may well make up for the lack of Cores.

 

Very interesting results. It is good to see finally someone could give some input on the issue.

 

20 fps to 30-40fps is quite a large difference indeed (50-100%), so I don't agree w.r.t. unreliability. Stutter is more difficult to account for, though. I've also read that stuttering in GTA 4 happens pretty much at random. On the whole, though, GTA 4 seems to be very dodgy indeed.

There has to be something relatively complicated behind the issues with GTA 4, because the difference can't be explained only in terms of computational power.

As for desktops, the differences in clock speed between quad-cores and dual-cores are much smaller there (as a percentage); although the higher clock speeds of desktop CPUs may be enough for GTA 4 to run fine, quads should still have a big advantage if your benchmarks are anything to go by.

 

How was your testing done? Did you pop out the 820QM and pop in the 620M or what?

 

Yes, I did exactly that. I wanted to know whether I would get better performance with a higher clocked Dual Core with PS2 emulation as well as other Dual Core applications including GameCube emulation but since GTA IV is my favourite game I also tested performance in that as well.

I also notice very similar results if I simply disable the affinity of my Cores leaving only two for GTA IV.

According to other 620m owners, games such as Empire Total War which is supposed to be very CPU intensive works perfectly fine. GTA IV appears to be a special case but it is good to have the option of being able to play it well.

With the PS2 and Gamecube emulation, I got exactly the same performance with the i7 820qm in Turbo Boost Dual Core mode as I did with the i7 820m. i also noticed the Dual Core always had a higher percentage of CPU usage in general.

It is strange that the Dual Core hits 80% usage in GTA IV while the Quad only hits 50%. I have a few theories that don't make sense lol.

Maybe it only needed 50% of the Quad's total computional power at any given moment to process all of the CPU intensive calculations and AI but to do the same on a Dual Core, perhaps it required almost double the CPU cycles due to lacking the extra Cores.

 

I think everyone on this thread should take a quick read of the last few pages of this thread. It explains hyperthreadings advantages/disadvantages very well.

 

The thing is, if 50% was enough then only two cores of the i7-820QM would do the job - that's 2 2GHz Nehalem cores, while the i7-620M offers 2 3.067GHz Nehalem cores.

There are other differences such as better memory bandwidth and latency in the i7-820QM, and more cache. However, most benchmarks show that neither of these have such a huge impact as your GTA 4 results would suggest.

I, too, cannot really say what causes this kind of behaviour in GTA 4, though I would love to find out exactly what goes on with this game.

 

GTA 4 is a very badly coded game. To use that as a bench is problematic. But so be it. It is what it is.

 

Yes it is a very badly coded game but it is also a very fun game to play and even though it is a bad port, it is perfectly playable on some laptops and very enjoyable, especially with the Mods. I wouldn't give it the time of day otherwise.

 

I only ever played GTA3 and a little bit of Vice City. Is it kinda just the same old same old?

Also it really must be a horrible port because depending on settings it can bring down 4+ Ghz desktop i7.

 

I'm running GTA IV perfectly fine on a 720QM at 1.73GHz (quad core turbo).

Flight Simulator X, on the other hand, could use a 4+ GHz desktop i7 to run at max graphics with a ton of addons. It pegs all 4 physical cores of my 720QM at 100% and stutters like crazy.

 

thanks for the link, Jkleiss!

 

It is more of the same but the increased realism and the fun you can have with the new and unpredictable physics engine gives it a new lease of life.

 

Flight Simulator X was designed for Dual Core and works better with a higher Clock rate and two Cores. The chances are that the multicore patch they released (SP2) causes less performance with mobile i7 because it ends up using 4 Cores in stead of two Turbo Boosted ones. Benchmarks have also shown it gets better performance in relation to clock rate as opposed to number of Cores as well.

Maybe the addons make this a different scenario due to the increased calculations needed after they are used though.

 

The framerate does go up if I force it to run on 2 cores with the affinity setting in task manager, but then it stops loading autogen (trees, houses, etc) entirely. Does anyone here run FSX with a 620M and would like to share their experiences?

 

I recently purchased FSX and am waiting on my 8gb of ram turning up (the 620M is already fitted now, easy upgrade). I will post back when the m15x is up and running (sometime this week).

 

What CPU did you have before the upgrade?

 

i3 330M, I didn't like the idea of paying £300 to upgrade to the 620M with Dell so I got it for £172 off ebay (and yes, it's the OEM rather than the ES).

 

I'm running a 320GB 7200 RPM WD Black at the moment but I generally don't run my VMs off of the internal HD. I have a separate, 3.5" 640GB WD Black that I connect via eSATA.

I'm currently only running 4GB of RAM, so that is definitely causing issues with VM performance as I'm only allocating up to 1500MB at a time to the VM, but I'm trying to figure out whether I'm going to notice a big improvement by upgrading to an i7-620M (2.67 GHz) based E6410 or if I should just upgrade the E6400 I have now which has a P9600 in it (also 2.67 GHz).

Upgrading the E6400 would entail purchasing 8GB of DDR2 ($300), an SSD ($350), and an HD caddy for my drive bay to put my current HD there ($50). So that's $700 to upgrade my E6400.

Other option is to get an E6410 from Dell Outlet for about $800 and spend the same money to upgrade it (except with DDR3 instead of DDR2) for a total of ~$1500. Question is whether $800 is worth it to get the new chassis -- will the performance justify it? Or should I just upgrade the E6400 and suck it up?

In all honesty, the 2.67 GHz P9600 E6400 has performed most admirably for me. It's only with the release of Visual Studio 2010 and SharePoint 2010 that it's starting to feel a bit underpowered. I'm really not sure if a 2.66 GHz i7-620M E6410 will make it that much better.

I would think that this would be an obvious choice for doing a comparison (2.67 GHz P9600 E6400 vs 2.67 GHz i7-620M E6410), but I haven't seen any comparisons to give me any good indication of the performance disparity between these two systems.

 

Keep in mind the i7 has turbo boost and runs at 3.06GHz with both cores loaded and 3.33GHz with one core loaded, so it's not really fair to use its base speed in the comparison.

 

At some point I'lll have access to both an M15x with an i7-620m, and a G51J with an i7-720qm. The comparison won't be perfect since the gpus will be a 5870 in the M15x, and a GTX 260m in the G51J, but it should still be a fun comparison on lower settings.

 

I would run perfmon.exe in both your VM's and your host to see where your bottleneck is, and decide from there. In addition to processor time, add memory pages/sec, and Avg Disk queue length if they aren't already there.

high cpu usage speaks for itself

memory pages/sec is pages per second to the hard disk. High numbers here show a need for more memory. This can also cause excessive drive activity as memory is paged to the disk.

a high disk queue length shows a bottleneck in the disk i/o


In my experience with VMWare and Hyper-V, VM's are limited in this order

1. disk I/o
2. memory
3. cpu

In fact, I run a cluster of VMWare servers, and all of the host machines have access to a 2.5tb 4gbps fiber channel SAN stuffed full of 450gb 15,000 RPM SAS drives, 32gb of memory (utilization around 24-28gb), but the CPU's are only around 40-50%.

 

Interesting feedback. I'm guessing that my performance issues are exactly as you postulated: the low RAM assigned to the VM is causing excessive disk paging.

It seems that it might be more economical to get the 8GB and SSD and forgo the chassis/CPU upgrade.

 

I can't say 100% that's the issue, but I run multiple VM's on my E4300 (SP9400 and 8gb memory) on an Indilinx 128gb SSD and it runs great. I don't do the same stuff you do, but I run Exchange, Zimbra (another email server), SQL 2008 etc and it does fine for a laptop, and my CPU is never over 50% for any period of time.

 

Some suggest that 50% CPU usage refers to the number of Cores the OS sees i.e 8 Cores and not 4 due to hyper threading so in that case 50% CPU usage may well refer to the use of 4 Cores meaning 2 of the Dual wouldn't be enough.

I enabled Rivatuners in game OSD to display Core usage and when the overall percentage was around 49 or 50%, I noticed many Cores out of the 8 displayed had very high CPU usage. if I had a way to disable hyper threading then it would be clearer.

 

have you tried the game with just 3 cores? This wild thinking (not sure if you've done it already) is just that the Xbox 360 is a triple core.

 

Yea I have tried that before and with three Cores, it is not as smooth or consistent but it is still perfectly playable and nowhere near as bad as performance was with 2 Cores. The extra Core seems to make up for their bad code efficiency I guess.

 

You're right about the Hyper-Threading, I think. Running wPrime with 2 threads, the CPU usage on my i5-430M is reported as 50%. Even if that's the case, it's not just a matter of 4 cores vs 2. 2 cores at 3.067GHz are quite fast, and are basically equal to the 4 cores in, say, the i7-720QM. (In terms of raw power they're behind by 13%, but even video encoding doesn't scale perfectly to four cores, and so this advantage disappears when you look at benchmarks).

You should try disabling Turbo Boost in your i7-820QM, since that would reduce it to 1.73Ghz and so it would perform similarly to an i7-720QM (although it's not quite the same since the 820QM has more cache).

Here's one suggestion I'll put forth - perhaps Hyper-Threading is actually what's responsible for the poor performance in GTA IV. Your figure of 50% for the i7-820QM probably means that the game only uses one logical core per physical core, so Hyper-Threading has no impact. On the other hand, 80% usage for the dual-core would suggest Hyper-Threading is used to some extent, but it could in fact be affecting performance negatively for the dual-core. Perhaps setting affinity to, say, Cores 0 and 2 would make GTA 4 run much better on the i7-620M? I've heard that affinity doesn't necessarily force it to only use those cores, though, so a proper way of disabling HT might be required.

 

Setting GTA IV to run on Cores 0 and 2 reduced performance by a few fps. At some points it was only running at 10 fps

All other games I tried perform perfectly with that CPU although it seems the i7 820qm does not lose out to the 3.06 Ghz Dual Core. I will be removing the i7 620M ready to be sold but I will benchmark Fsx to see if it performs better. I expect it will.

Edit: Fsx didn't perform much differently to how it did on the Quad but I think that is due to the multicore patch. However, without that patch (confusing Turbo Boost)Turbo Boost of the Quad would also be closer to the Dual anyway.

 

Sure, but the i7-820QM is significantly more expensive than the i7-620M, so it's not the fairest of comparisons. GTA IV is an anomaly at best, and I still don't see an explanation for why it behaves the way it does.

For gaming, you've seen that the i7-620M is the equal of the i7-820QM for all but one game. Plus, the faster Core i5s aren't far behind the i7-620M either, and they're even cheaper.

 

Well I am assuming that there won't be any other games like GTA IV in terms of being an anomoly, however, until BFBC2 was patched after release, it was a very similar situation too so you never know. Console Ports are notorious for being poorly optimised so I take that into consideration as well.


The i7 820qm is more expensive but the i7 720qm is much closer in cost (in the Uk anyway).

With the vast majority of games being more GPU dependent, I would expect most current CPU's to be fine though.

 

So here is my dilemma. I plan to buy a dell studio 15 here in Australia and here are my processor choices with upgrade costs.

i5-540M [Included in Price]
i7-620M [add $125.40]
i7-740QM [add $20.90]
i7-840QM [add $449.90]

I plan to use it for code development with servers in VMs, and video editing/transcoding. (btw I'm not a gamer)

Here are some Sony Vegas benchmarks I found here -> i5 or i7 for video rendering in Vegas 9? - Notebooks - Whirlpool Forums

Sony Vegas HD 1080p MP4 9 (4minute 55sec Music Video) Rendering:
Intel Core i7-720QM 1.6Ghz: 6Minutes and 14 Seconds
Intel Core i5-540M 2.53Ghz: 9Minutes and 03 Seconds

My choice is between the i5-540M and the i7-740QM as I can't justify the extra $125.40 for the i7-620M.

I was going to go for the i5 because I have heard of heat/stability/throttling issues with the clarksfield i7s. But based on the above benchmarks, for the extra $21, it seems to be a good idea to go with the i7-740QM, but this thread seems very pro arrandale.

Anyone have any thoughts they would like to share that might help me decide?

 

i edit 720p video using my 720qm using both power director 8 and ppro cs5. multicore processors make a lot of difference in these programs in both editing and rendering.
for 1080p videos however i would recommend at least an 820qm.
almost everyoe ive spoken to who does video editing will pick a quad over a dual core, even if the quad was a few mhz slower.
in short get the 740qm

 

The 740QM will definitely be faster for video editing than the i5-540M; it's a great upgrade for $20.

 

@ Trottel, @ LackOfCheese

I was reading this thread and was wondering whether you had seen it yet. It poses some interesting issues regarding emulation and hyper threading.

I would be very interested in your input on this one.