- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Denunciar
s.brueckner,
Thank you for your reply. I have long been a PC extreme geek... window mods and water cooling, enough said? I did check the thread you referenced. It would be nice to have a time frame for those reports about lower performance Xeon CPUs platforms. As I recall, 2-3 years ago there was a significant speed delta between the two, both in CPU and memory clock. Today, the Xeon's are often clocked very similarly in CPU and memory interface, so I have reduced this concerns. This time around, I simply do not have the time, so I wound up at Dell. Xeon 3.7GHz, 16gig, Quadro M4000, SSD. I'll share some of what I've seen just for reference.
I've been using FeatureCAM for... a long time, over 15 years as I recall, but my oldest FM file currently on file is around 2/2002, guess that job's never coming back. The software and hardware has changed significantly over the years, mostly for the better, but over the last 2-3 years I have not seen much improvement. IMO, Windows 7 Pro and 32bit FeatureCAM was a great combination.
In January 2016 I was forced to migrate our internal network structure from Windows 2003 SBS w/ a mix of XP and Windows 7 to Windows 2012 Essentials, Office 365 w/ hosted exchange server and Windows 10. It's been a horrible experience.
FeatureCAM has forced users into the 64 bit versions with near zero user experience improvements. I have seen no improvement in speed from 32 bit to 64 bit even when working with large files containing STLs. Much of the internal code still appears bound to a single core, making little use of the multi core CPUs. In the Windows 10 environment file conversion from 32 bit to 64 bit takes a MASSIVE hit. Massive is defined as 30 minute file convert times. (hint: Support did share, if you save a 32 bit file as "compressed" it will improve import performance. 30 minutes down to 5). The heartache experienced in migration encouraged me to test FeatureCAM on many platforms.
v22 64 bit has been tested on: Phenom II, i3, i5, i7, GPUs ranging from integrated intel to GT9800, Radeon 6900, to GTX 670.
The 1st major change which was 100% demanded was migrating from ODBC connectivity for the shared tool database to SQL. v22 was happy with this one change and I'm grateful that FeatureCAM implemented SQL. Learning curve for installing SQL Express was about 1 day. Maintaining backward compatibility is requiring additional effort and is not as easy.
FeatureCAM performance saw little difference on any platform... until v23 Windows 10 64bit and tombstone simulation. Loading the tombstone fixture with 4 part files which used STLs as input material for 'second side' type operations caused a start simulation wait time of several minutes (over 5 minutes). This is the time elapsed between pressing play in the solid simulation and beginning to see the 1st tool cut.
The PC that was running FeatureCAM was a work horse. Back in 2010 I discovered that old school compatibility for VM machines hosting DOS were supported perfectly by AMD CPUs, P4 and Core 2 broke some basic rules which prohibited support of old Dox programs. I built a Phenom II X5 955, 4Gig, Gforce 9800GT, 7200RPM HD to support some of our legacy work requirements. Upgrading this PC to the Xeon has reduced the time to about 2 minutes. My rig is an i7 950 w/ 12gig. Wait time on my rig is also 2 minutes. On both the Xeon and i7 pressing play a second time is less than a 1 minute wait.
My initial impression is a new PC platform running Windows 10 and FeatureCAM 2017 does result in a significant performance improvement on newer platforms. Moreover, there seems to be little difference between the i7 and current generation Xeon platform performance. Our investment in the Quadro GPU is more of a safe play with little real world improvement just yet. I believe this is a significant statement because new workstation class builds from major manufacturers tend toward Xeon based systems.
-Peter.