1 4 5 6 7 8 9 Previous Next 132 Replies Latest reply on Jun 26, 2011 1:07 PM by slehman805 Go to original post
      • 75. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
        woodmeister Enthusiast

        First, I am an electrical engineer that develops both embedded hardware and uP code

        and also do FPGA design.  Also, I am not a gamer!!!  I use windows VM's because

        most of the tools I use are only available for the windows platform.

         

        Therefor, after hearing so much flap about Parallels, I decided to give it a

        shot.  Was hesitant as my first experience with them was a nightmare on a PC back

        around version 2 which prompted me to move to Vmware Workstation and once

        the Intel platform platform matured on the MAC I moved to Fusion.  I had been very

        happy with Fusion's performance up til Version 3 and Windows 7.  For the most

        part though I had found no real issues other than slower performance on various

        aspects of the product.  I also must state at this point that I am a Fusion advocate.

         

        Well, my experience working with Parallels the last months as been a difference of night and

        day.

         

        First, I should note the platforms that I used.  One is a Macbook4.1 with a 2.4GHz processor

        and 4G RAM,  the other, an IMAC9.1 with a 2.66GHz processor and 4G of RAM.

         

        The the Windows7 VM used in Parallels was imported from the Fusion VM so both

        windows installs were identical.

         

        Some examples:

         

        1.  A free schematic capture and PCB layout program called PCB artist:

             Fusion -  Graphic redraws were so slow and jerky so as to render the schematic

                           capture portion of the program useless.  Never got to layout.

             Parallels - All actions in both the schematic capture and layout were very smooth

                            and fluid and made the package a joy to use.

        2.  USB performance

             1.  File copying to and from USB sticks

                  Again, Parallels was much faster in the respect.

             2.  Function of programmers

                  I have used the Cypress Miniprog3, Microchip PICkit2 and ICD3 and the download,

                  upload and debugging response times are noticeably faster using Parallels.

             3.  Linksystems DSO-8502 Oscilloscope had much smoother screen draw performance

                 with no occasional stalls with Parallels as was the case with Fusion.

        3.  LTspice from Linear Technology

             When doing a transient analysis on a fairly complicated circuit with a long duration,

             results were returned in nearly in half the time with Parallels.   A note here is that

             this was based on a setup where the waveform was being drawn as the simulation

             progressed.  In a case where the simulation was done first then the results plotted,

             the calculation time was noticeably faster in Parallels but not by a huge amount

             but drawing f the subsequent plot was a lot faster.

         

        I could continue on with other examples but it seems to me Parallels has faster graphics,

        performs faster with applications that require a lot of disk i/o, and the USB interface

        performance seems much better.

         

        As far as CPU intensive tasks, I coud say that I can notice some speed improvement

        in Parallels but not enough to be significant.

         

        Now, I am not advocating that everyone should drop Fusion, as a matter of fact I am a

        strong supporter of the Fusion product.  But the Fusion product has fallen behind Parallels

        by a pretty significant amount and have a lot of catch up to do! 

         

        Hopefully, when 3.1 matures and is released as final, it will catch up.  At which point,

        I will return to using Fusion full time.

         

        P.S.  I haven't tried comparing any of my Linux machines as I just haven't had the time and

                also, importing them always seems to be an adventure if they import at all.

        • 76. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
          dicen Novice
          GlennyG wrote:

           

           

           

          What really sucks is a bunch of selfish gamers driving development of 

          what is ostensibly a business tool.

          </div>

          You think the GPU is only used by games? This is from the 3.1 beta thread (no it doesn't fix these issues either).

           

           

          Scrolling and selecting text with Aero enabled in VS2008 is really slow (it takes almost a second to select 15 lines of text). It's unusable slow. I have a Core i7 iMac 27" 8GB and ATI Radeon HD 4850. I have reserved 4GB for the VM. With Aero disabled scrolling is a lot better than my XP VM, so you guys made progress for Win7. I am not sure on what kind of Macs you guys test Windows 7, but graphics performance is horrible. If the experience on my Mac is as it should be, then VMWare 3 doesn't support Aero.

           

           

          Yes, that is Visual Studio 2008 an ostensibly business tool. What really irks me here is that VMware markets this product as one that supports DirectX 9 and Aero. That is obviously an out right lie and now numerious sites on the net have called them on it. If it can't even render the graphics right it doesn't support it. If they were honest about it that would be one thing, but they are not. Very unethical.

           

           

          http://www.engadget.com/2010/03/16/mac-virtualization-face-off-vmware-fusion-3-vs-parallels-deskt/

           

           

           

           

           

          • 77. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
            lugesm2 Novice

            woodmeister . . .

             

            You stated,

             

            "The the Windows7 VM used in Parallels was imported from the Fusion VM so both

            windows installs were identical."

             

            Are you saying that one can simply copy the VM in Fusion and paste it into the proper folder in Parallels?

            Or, is the "importing" a more complex process ?

             

            Thanks.

            • 78. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
              lugesm2 Novice

              Another question:

               

              Is there any conflict with installing Parallels 5 on the same Mac where Fusion 3 is already installed ?

               

              I should note that I have been a LOYAL user of Fusion 1 since I first got my Mac, and I am VERY reluctant to change over.

              Perhaps testing both Fusion 3.01 and Parallels 5 on the same machine would provide good comparison data.

              • 79. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                Species8472 Novice

                AFAIK you can install Fusion as well as Parallels, but you can't run a Fusion VM side by side with a Parallels VM.

                 

                Am 17.03.2010 um 13:52 schrieb lugesm2:

                 

                ,

                 

                A new message was posted in the thread "Fusion 3.0.1 heavily beaten by Parallels 5 in performance":

                 

                http://communities.vmware.com/message/1495757#1495757

                 

                Author  : lugesm2

                Profile : http://communities.vmware.com/people/lugesm2

                Message:

                 

                • 80. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                  HPReg Expert
                  VMware Employees

                  I just feel we need to keep this thread alive to remind fusion team the customers

                  are still waiting for performance improvements.

                   

                  There is no need for this. It just creates unnecessary noise in our mailboxes. The Fusion team is well aware of the performance problem, and we are actively hiring in that area (see http://tinyurl.com/ycafxra ).

                   

                  Heres yet another fresh benchmark:

                  http://www.engadget.com/2010/03/16/mac-virtualization-face-off-vmware-fusion-3-vs-parallels-deskt/

                   

                  This is a re-hash of the same issues. Looks like journalists have found an easy target.

                  • 81. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                    bytekeeper Novice

                    Sorry HPreg. I have no doubts you guys are aware of the problem and willing/going to fix it sooner or later. However, it is clear that you don't put enough effort into this. Look, the problem is there for quite long time now and still you are only hiring someone to cope with it. So thats another half on an year at best until its addressed. I'm sure || will come out with 6.0 by that time. I think it is just unacceptable for the company of this class! You are not a bunch of coders from garage office after all.

                     

                    I'm trying, trying hard to not lose the confidence in fusion but it is getting more complicated with every month. I'm giving up if 3.1 is out with same level of "no one cares about" updates from 3.0.1-3.0.2.

                     

                    Sorry for the rant, shutting up till 3.1.

                    • 82. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                      dicen Novice
                      HPReg wrote:

                       

                       

                       


                      I just feel we need to keep this thread alive to remind fusion team the customers
                      are still waiting for performance improvements.

                      There is no need for this. It just creates unnecessary noise in our mailboxes. The Fusion team is well aware of the performance problem, and we are actively hiring in that area (see http://tinyurl.com/ycafxra ).

                       

                      Heres yet another fresh benchmark:

                      http://www.engadget.com/2010/03/16/mac-virtualization-face-off-vmware-fusion-3-vs-parallels-deskt/

                      </div>This is a re-hash of the same issues. Looks like journalists have found an easy target.

                      </div>

                       

                       

                       

                      They are running 3DMark and are getting black screens on Fusion! You don't support 3DMark? But, how can that be when you say you support DirectX 9 SM3? I am very confused here. 3DMark is a very common DirectX 9 app that is used to verify graphics card/driver functionality and well as performance.

                       

                       

                      This VMware Fusion features web page has the following info http://communities.vmware.com/docs/DOC-10957

                       

                      3D Graphics Now Supports More Applications and Games

                      • Support for Windows Aero with WDDM driver for Windows 7 and Windows
                        Vista

                      • DirectX 9.0c with Shader Model 3 support for Windows virtual
                        machines

                      • OpenGL 2.1 support for Windows XP virtual machines

                       

                       

                       

                      You're hiring? But, you already said you included the functionality above. Isn't VMware owned by EMC now? A multi-billion dollar IT company. Why do you need to hire people when a little bitty company like Parallels already has it working? How is that possible? Why is it that they have the people and you don't? Hmmm... something is wrong with this picture. You have revenue of over a billion dollars and yet a small little company with 750 employees and only $100 million a year in revenue has it working perfectly.

                       

                       

                      VMware

                       

                      Revenue\

                      [US $\

                      http://en.wikipedia.org/wiki/United_States_dollar\

                      United

                                     States dollar] 1.9 Billion

                      Employees\

                      7,100

                      [Parent\

                      http://en.wikipedia.org/wiki/Holding_company\

                      Holding

                                     company]

                      [EMC Corporation\

                      http://en.wikipedia.org/wiki/EMC_Corporation\

                      EMC

                                     Corporation]

                       

                      Parallels

                       

                      Revenue\

                      [US$\

                      http://en.wikipedia.org/wiki/United_States_dollar\

                      United

                                     States dollar] 100 million (2008) ^[[2]\

                      http://en.wikipedia.org/wiki/Parallels,_Inc.#cite_note-1]^

                      Employees\

                      ~750 (as of Jun 2009\)

                       

                       

                       

                       

                      $1.9 Billion / $100 million = 19x as large

                       

                       

                      7100 / 750 = 9.46x as large

                       

                       

                      Well, he said they were hiring folks so we are not going to get this fixed anytime soon. Major brain drain. I find it very interesting that Parallels also has a number of server virtualization products as well but nearly 10x less employees.

                       

                       

                       

                       

                       

                      Check out EMC. VMware should have all the geniuses the world over. Something is definitely very wrong with this picture. I wonder what?

                       

                      Revenue\

                      [$\

                      http://en.wikipedia.org/wiki/United_States_dollar\

                      United

                                     States dollar] 14.876 billion (2008)^[[1]\

                      http://en.wikipedia.org/wiki/EMC_Corporation#cite_note-financialtables-0]^

                      Operating income\

                      [$\

                      http://en.wikipedia.org/wiki/United_States_dollar\

                      United

                                     States dollar] 1.568 billion (2008)^[[1]\

                      http://en.wikipedia.org/wiki/EMC_Corporation#cite_note-financialtables-0]^

                      [Net

                                     income\

                      http://en.wikipedia.org/wiki/Net_income\

                      Net income]

                      $\ 1.345 billion (2008)^[[1]\

                      http://en.wikipedia.org/wiki/EMC_Corporation#cite_note-financialtables-0]^

                      Total assets\

                      [$\

                      http://en.wikipedia.org/wiki/United_States_dollar\

                      United

                                     States dollar] 23.874 billion (2008)^[[1]\

                      http://en.wikipedia.org/wiki/EMC_Corporation#cite_note-financialtables-0]^

                      [Total equity\

                      http://en.wikipedia.org/wiki/Ownership_equity\

                      Ownership

                                     equity]

                      [$\

                      http://en.wikipedia.org/wiki/United_States_dollar\

                      United

                                     States dollar] 13.041 billion (2008)^[[1]\

                      http://en.wikipedia.org/wiki/EMC_Corporation#cite_note-financialtables-0]^

                      Employees\

                      42,100 (March 2009)

                       

                       

                       

                       

                      • 83. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                        brainee28 Novice

                        This is getting way out of hand here.

                         

                        HP Reg has said they're aware of it, and they're trying to work on it, including hiring outside help to work on the performance portion.

                         

                        Ever been in a Parallels forum? They tell you absolutely nothing. No response. So most of you should be lucky that you're getting one from VMWare.

                        Parallels hasn't figured out how to effectively talk to their users. It adds a layer of frustration to an already tense situation.

                         

                        @HPReg: I think it's fairly obvious due to the length of the thread that Fusion 3 has performance issues. It's also obvious that the customers that purchased Fusion 3 did so with the understanding that performance would be at least equivalent to version 2, if not better.

                         

                        This doesn't sit well when someone buys an upgrade, thinking it's better, and finding it goes slower. Whether it's anecdotal or statistical, the perception is that it was supposed to be better. I personally have noticed this problem. I alleviated it somewhat with aligning my partition, and actually changing from 1 processor to 2 seemed to make it run a lot smoother, along with my Mac apps.

                         

                        Often times, companies disregard the end user and just build what they think is the product people want. Why some of these people are posting is that they don't want VMWare to sidestep the issue, ignore the issue or pay lip service to get past the problem. Now I don't think that's yours or VMWare's intention; they just want to ensure that's not the case.

                         

                        I guess what I'm trying to say is this: Address the performance issues with a sense of urgency here. This seems to be important, so don't blow it off.

                        I understand that Fusion developers must be tired of hearing people complain about this, but this directly sells the product. Performance along with stability, flexibility, and scalability are extremely important, and some people are getting the impression that the Fusion developers didn't think so when they released version 3.

                        • 84. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                          dicen Novice
                          brainee28 wrote:

                           

                          This is getting way out of hand here.

                           

                          HP Reg has said they're aware of it, and they're trying to work on it, including hiring outside help to work on the performance portion.

                           

                          Ever been in a Parallels forum? They tell you absolutely nothing. No response. So most of you should be lucky that you're getting one from VMWare.

                          Parallels hasn't figured out how to effectively talk to their users. It adds a layer of frustration to an already tense situation.

                           

                          @HPReg: I think it's fairly obvious due to the length of the thread that Fusion 3 has performance issues. It's also obvious that the customers that purchased Fusion 3 did so with the understanding that performance would be at least equivalent to version 2, if not better.

                           

                          This doesn't sit well when someone buys an upgrade, thinking it's better, and finding it goes slower. Whether it's anecdotal or statistical, the perception is that it was supposed to be better. I personally have noticed this problem. I alleviated it somewhat with aligning my partition, and actually changing from 1 processor to 2 seemed to make it run a lot smoother, along with my Mac apps.

                           

                          Often times, companies disregard the end user and just build what they think is the product people want. Why some of these people are posting is that they don't want VMWare to sidestep the issue, ignore the issue or pay lip service to get past the problem. Now I don't think that's yours or VMWare's intention; they just want to ensure that's not the case.

                           

                          I guess what I'm trying to say is this: Address the performance issues with a sense of urgency here. This seems to be important, so don't blow it off.

                          I understand that Fusion developers must be tired of hearing people complain about this, but this directly sells the product. Performance along with stability, flexibility, and scalability are extremely important, and some people are getting the impression that the Fusion developers didn't think so when they released version 3.

                          </div>

                          I didn't say I was irked about the lack of performance. I am irked about the lack of functionality. They are selling a product that says it supports DirectX 9 SM3. That is plainly false if it can't even run 3DMark without black screens and numerious graphical errors. Being slow is one thing, being so broken that it displays a black screen is another. This is false advertising plain and simple.

                          • 85. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                            woodmeister Enthusiast

                            Parallels will "import" the virtual machine into it's own format and when

                            the VM is started it will install it's own tools.  Also note, there will be

                            an activation required after a three day period with Windows7. 

                             

                            P.S.  I would suggest backing up your VM just in case you do

                            something wrong.

                            • 86. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                              HPReg Expert
                              VMware Employees

                              3DMark: Yes, we are a incorrect/slow on some Mac models which came out after Fusion 3.0 came out (such as the ones with Core i7 CPUs). Expect a lot of improvements on that front in the next Fusion 3.1 refresh (not the one that came out a week ago) when running on a Mac OS 10.6.3 host.

                               

                              Company size: You are comparing apples and oranges. Yes, VMware has a much larger headcount and revenue that Parallels. But keep in mind that at VMware, only a fraction of the company works on the desktop products (which is in relation to the revenue these products bring), and of that, only a fraction works on Fusion. Maybe you won't believe me, but I estimate that the Fusion team is about 10x less numerous than the Parallels team. While we do need to improve on the perf front, and we are working on it, we have done the best we could so far given our resources. That being said, we do have people working on Fusion perf right now, and we are hiring more.

                              • 87. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                                blopzy Lurker
                                HPReg wrote:

                                Regis,

                                 

                                While you don't necessarily need to provide excuses, you also shouldn't be trying  to disguise here.

                                 

                                 

                                 

                                3DMark: Yes, we are a incorrect/slow on some Mac models which came out after Fusion 3.0 came out (such as the ones with Core i7 CPUs). Expect a lot of improvements on that front in the next Fusion 3.1 refresh (not the one that came out a week ago) when running on a Mac OS 10.6.3 host.

                                The notorious Mactech video shows Fusion failing to display 3DMark on unibody MacBook Pro, which were available since 2008, long before Fusion 3. So, the problem is not only with the latest hardware as you're trying to present.


                                On the other note, OS X 10.6.3 isn't out yet. Why not simply support current 10.5.8 and 10.6.2?



                                Company size: You are comparing apples and oranges. Yes, VMware has a much larger headcount and revenue that Parallels. But keep in mind that at VMware, only a fraction of the company works on the desktop products (which is in relation to the revenue these products bring), and of that, only a fraction works on Fusion.

                                Are you saying that Fusion shares no code with ESX and Workstation? As you stated in another thread, at least the VM monitor and guest tools (like graphics drivers) are shared. And I bet, much more code is common for VMware virtualization products. This effectively makes the official Fusion team size irrelevant because it uses many common resources, right?



                                Maybe you won't believe me, but I estimate that the Fusion team is about 10x less numerous than the Parallels team.

                                This is just unfair to Parallels and you most probably can't prove that claim. Parallels also has other products besides Desktop fro Mac, and they are free to balance their resources as they wish. VMware has 10x more manpower, so the original posters' point is still valid.


                                I recall VMware marketing used the small Fusion team size to demonstrate advantage over the competition, and now you're using it in quite the opposite sense.



                                While we do need to improve on the perf front, and we are working on it, we have done the best we could so far given our resources. That being said, we do have people working on Fusion perf right now, and we are hiring more.

                                 

                                 

                                 

                                 

                                Actually my personal complaint is only with missing OpenGL for Linux support. VMware has basically bought full Linux OpenGL code and team (Tungsten graphics) and is still unable to leverage that. And after all you still talking about not having enough resources...

                                • 88. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                                  HPReg Expert
                                  VMware Employees

                                  you also shouldn't be trying to disguise here.

                                   

                                  It frustrates me a lot to answer in this forum, because whatever I say is necessarily interpreted the wrong way by somebody

                                   

                                  On the other note, OS X 10.6.3 isn't out yet. Why not simply support current 10.5.8 and 10.6.2?

                                   

                                  We made a choice very early on to virtualize the GPU by using the host GPU, and not do it in software. The competition made a different choice, basing their code mostly on Wine/Crossover.

                                   

                                  So far, the competition has been more correct and faster for 3D because the OpenGL stack on Mac OS is ahem... subpar compared to Windows/Linux, mostly because Apple only optimizes the paths they use. When the path that is used is that of a Windows driver, their drivers suck. But things are changing, and Apple will be providing much better OpenGL drivers in 10.6.3. That is why we care so much about 10.6.3.

                                   

                                  We believe our architecture is the right one for the long-term. For example, using the competitior's architecture, you cannot correctly suspend/resume a VM while it is doing something in 3D (game, or os window manager). Our architecture can. So time will tell who is right.

                                   

                                  Are you saying that Fusion shares no code with ESX and Workstation? As you stated in another thread, at least the VM monitor and guest tools (like graphics drivers) are shared. And I bet, much more code is common for VMware virtualization products.

                                   

                                  Yes

                                   

                                  This effectively makes the official Fusion team size irrelevant because it uses many common resources, right?

                                   

                                  No. A lot of the optimizations happen in the device backend, which are specific to the host OS.

                                   

                                  This is just unfair to Parallels and you most probably can't prove that claim. Parallels also has other products besides Desktop fro Mac, and they are free to balance their resources as they wish.

                                   

                                  I don't think my claim is far from reality. The reality is that Parallels sells little out of their Mac product. This is their lifeline, so they put most of their resources on it. That is not the case of VMware. While they are mostly focusing on their Mac product, we are mostly focusing on our datacenter products.

                                   

                                  I recall VMware marketing used the small Fusion team size to demonstrate advantage over the competition, and now you're using it in quite the opposite sense.

                                   

                                  I'm an engineer, I cannot answer for what the marketing team does

                                   

                                  Actually my personal complaint is only with missing OpenGL for Linux support. VMware has basically bought full Linux OpenGL code and team (Tungsten graphics) and is still unable to leverage that.

                                   

                                  Have you tried the Linux tools which come with Fusion 3.1 beta and WS 7.1 beta? They have OpenGL support if you provide a non-default config option to the installer.

                                  • 89. Re: Fusion 3.0.1 heavily beaten by Parallels 5 in performance
                                    ae3265 Novice

                                     

                                    HPReg:

                                     

                                     

                                    As a customer it frustrates me a lot that in the time you'e spent defending the state of things, you could have been looking into the headless mode that got broke in 3, or at least bothered to respond to my last query about it.

                                     

                                     

                                    Actually, as I use Fusion more from a server perspective, I can deal with the video issues.  Lack of a complete and properly functioning headless mode, which was present in 2 and thoroughly broken in 3, is a mch bigger issue and led to my use of competing products.

                                     

                                     

                                    1 4 5 6 7 8 9 Previous Next