nVidia Quadro FX580 vs. FX1800

[gleonard]

Pro/ADMIN's,

We're interested in saving a little money on new workstations. We're looking at the above mentioned cards and wanted to know the WF4 performance impact of going with the cheaper card. To study this, I cooked up my own graphics focused benchmark by plagiarizing much of Olaf's OCUSB5 Benchmark, which uses models specific to the work we do here.

Surprisingly, the numbers came out just about equal for both cards in every category. Does that seem right? Anyone else checked out the FX580?

Thanks,
--
George Leonard,
Space Systems/Loral
T:408-372-7763
F:650-852-7170


This thread is inactive and closed by the PTC Community Management Team. If you would like to provide a reply and re-open this thread, please notify the moderator and reference the thread. You may also use "Start a topic" button to ask a new question. Please be sure to include what version of the PTC product you are using so another community member knowledgeable about your version may be able to assist.

[StephenW]

Just Monday I got a new graphics card, not the same as yours, but it's very interesting and I would like comments from the rest of peanut gallery as well.
Went from FX1700 (bought with the HPxw6600 about 18 months ago) to FX4800. Overall, the benchmark showed a 25% increase in speed. The only thing different was the video card. If the spreadsheet doesn't go through, let me know and I will send it a different way



OCUS Benchmark v5-64bit







Machine1

Machine1





Name









Make









Model









Processor









RAM









Graphics

FX 1700

FX 4800





Operating System









Version









Test

Result1

Result2

Diff

1

Retrieve generic assembly (CP)

81

64

-20.99%

2

Retrieve assembly 1 (CP)

93

72

-22.58%

3

Retrieve assembly 2 (CP)

107

82

-23.36%

4

Retrieve assembly 3 (CP)

123

93

-24.39%

5

Retrieve assembly 4 (CP)

139

104

-25.18%

6

Retrieve assembly 5 (CP)

156

116

-25.64%

7

Retrieve assembly 6 (CP)

174

126

-27.59%

8

Retrieve assembly 7 (CP)

188

138

-26.60%

9

Retrieve assembly 8 (CP)

205

149

-27.32%

10

Retrieve assembly 9 (CP)

222

161

-27.48%

11

Retrieve assembly 10 (CP)

239

172

-28.03%

12

Retrieve largest assembly (GR)

21

10

-52.38%

13

60 wireframe view redraws (GR)

50

45

-10.00%

14

24 wireframe view redraws with DATUMS on (GR)

182

77

-57.69%

15

1 hidden view redraws (GR)

113

81

-28.32%

16

1 Fast HLR activations (CP)

69

42

-39.13%

17

50 hidden view redraws with Fast HLR (GR)

86

80

-6.98%

18

1 shaded mouse spins (GR)

92

71

-22.83%

19

100 shaded view redraws (GR)

86

64

-25.58%

20

80 realtime rendered redraws (GR)

85

92

8.24%

21

3 shade calculations (CP)

116

58

-50.00%

22

6 wireframe mouse zooms (GR)

114

121

6.14%

23

initiate advanced shaded mode (CP)

77

47

-38.96%

24

4 advanced shaded mouse zooms (GR)

148

169

14.19%

25

16 very advanced shaded spins (GR)

80

50

-37.50%

26

1 save jpeg (CP)

43

32

-25.58%

27

end advanced shaded mode (CP)

59

34

-42.37%

28

6 screen translates (GR)

112

82

-26.79%

29

6 automatic regenerates (CP)

149

43

-71.14%

30

25 Perspective views (GR)

73

94

28.77%

31

10 Perspective zooms (GR)

92

123

33.70%

32

1 mass prop calculations (CP)

61

58

-4.92%

33

1 global interference checks (CP)

75

50

-33.33%

34

1 IGES exports (CP+DI)

264

182

-31.06%

35

1 drawing creations (CP)

58

55

-5.17%

36

150 dimensions show all (GR)

198

110

-44.44%

37

1 regen views HIDDEN LINE (CP)

146

110

-24.66%

38

1 regen views NO HIDDEN (CP)

135

102

-24.44%

39

1 PostScript file creations (CP+DI)

164

122

-25.61%

40

1 DXF File creations (CP+DI)

179

128

-28.49%

41

Erase all from memory (MEM)

15

10

-33.33%













Total

4869

3619

-25.67%



Graphics

1532

1269

-17.17%



CPU

3322

2340

-29.56%



Disk

607

432

-28.83%

[dgschaefer]

My last workstation had an FX500, which I assume is the FX580's
ancestor, and I was not happy with the performance. This one has an
FX3500 and I have no complaints.

I wish I could remember the specific issues I had with the FX500, but it
was several years ago. I just remember counting the days 'till that new
PC arrived. (I bet a forum search for 'FX500' might find my old posts.)

The advice I remember getting was to go with the upper end of the mid
range cards. High end, cutting edge cards come with a price penalty due
to their new technology, low end cards can't keep up. The mid range
cards are a good combination of decent price and good performance.

Doug Schaefer

[dgallup]

I believe there was another post (maybe on MCAD Central) about the FX1700 being a particularly poor performing card. It's model number seems to be completely misleading as to where it actually ranks in performance.

[jcrockett-2]

I agree with Doug's suggestion that mid-range Quadro's are fine for ProE (if your budget allows), and relatively future-proof for non-ProE applications. I also agree that the high-end cards are a complete waste of money! As you probably already know, the most important factor in ProE workstation performance is cpu single-thread horsepower.

Contrary to Intel's (and virtually all the OEM's) hype, Quad-core cpu is a total waste for ProEngineer (now and thru Wildfire 6) as Part regeneration is not multi-threaded (and won't be through at least Wildfire6). The current ProEngineer cpu champ is the Intel 670 (dual-core Nehalem @ 3.46 ghz, 3.72 ghz in Turbo mode). The only first-tier OEM workstation products that offer this cpu right now (as far as I know) are the Dell T1500 and the HP z200. The funny thing is that both Dell and HP market these as "low-end" systems.

It has always been my experience that on a single monitor, with the highest display resolution settings, that all Quadro graphics cards run ProEngineer about the same (see example below). The 5xx series Quadro has usually been my recommendation, and I have never witnessed a ProEngineer problem whose root cause was the 5xx series card.

I have discussed this issue directly with tech folks inside PTC, Nvidia, and HP, and they are aware of the lack of ProE performance delta across their entire graphics product line. I have seen no data nor heard any claim that the lack of performance delta changes with Wildfire 4 or 5.

Dual display with brutal processes running on both screens at the same time might be a different story, but I don't know any ProE users that actually do that. We have plenty of folks with dual monitors (and 5xx cards) that are doing just fine... but they are only doing graphics-intensive (> 5 million polygons) stuff on one monitor at a time.

I have been running Radar and Antenna Top Assemblies on a HP z200 with a Quadro FX 380 for the last few months with no ill effect.