We picked up this PR from the GeIL site:

Veloce, the Italian word for “speed”, hence the name of GeIL’s 4th generation top of the line EVO DDR3 Hardcore Gaming Memory - EVO VELOCE.

EVO VELOCE is equipped with the latest Veloce heat-spreader designed with GeIL’s MTCD - Maximum Thermal Conduction & Dissipation technology for superior cooling efficiency. The standard EVO VELOCE will come with Metallic Hot-rod Red heat-spreader while selective top end specifications receive the limited Frost White finish treatment.

Speed of up to 3000MHz plus, EVO VELOCE is designed for all of the latest motherboard chipsets for maximum performance and compatibility, including the Intel Z77 platform.

EVO VELOCE is the new DDR3 performance benchmark that redefines speed.

The Veloce series exist in several different versions, available via your favourite online shop, being it a dual or a quad channel kit. The visual difference is obtained the heatspreader colour, either choose between Frost White or Hot Rod red.

 

While being spoiled with stylish heatspreader designs alike the G.SKill TridentX or the latest Corsair Platinum, GeIL opted to slightly restyle their EVO Corsa heatspreader. It's all a matter of personal taste if you like it or not. If the performance is there, most won't really mind how they look. Note that the heatspreader design is not a low profile version, if needed GeIL offers those in their EVO Leggera range

 

 

 

 

This 2133C10 is Hynix RAM based, the heatspreaders are solidly attached by very sticky thermal tape.

 

 

 

Below are the different versions of this Veloce kit, note that just the the Dual Channel kits are listed. Quad channel versions are also available in all sorts of flavours:

 

The voltage anomaly is a bit confusing, our 2133C10 kit is rated according the GeIL website and SPD at 1.65Vdimm. Yet the sticker on our DIMMs clearly mention 1.5Vdimm. We tested at 1.5Vdimm and the sticks worked flawless during the HCI Memtest.

 

As usual we test the DIMMs at the box rated speed of 2133MHz, but also at higher memory dividers speeds. We then try to find the tightest main timings possible. This usually takes some hours of Memtesting to see if the DIMMs are holding up with the selected timings. Our main test program to solely test the RAMs is HCI MemTest Deluxe. It can be run from within Windows and at multiple instances, to ensure we cover all the free RAM you got. For 16GB I usually test up to 13GB to leave 3GB free for Windows 7 to play with.

 

 

The test setup comprises of the following hardware :

 

Asus Maximus V Gene motherboard 0078 bios

Intel i7-3770K CPU at 4.5Ghz (45 x 100)

Asus Nvidia GTX480 VGA

Western Digital 1Tb Caviar Green HDD

Windows 7 64bit Professional SP1 & fully patched

Corsair 1200W AX series PSU.

 

The results of the Belgian jury are :

1600MHz : CL9-9-9-27 2T 1.5Vdimm: taken as reference as most systems are sold with this ram speed

2133MHz : CL10-11-11-30 2T 1.5Vdimm (box rated VELOCE speed)

2400MHz : CL10-12-12-30 2T 1.65Vdimm (OC )

2600MHZ : CL11-13-13-30 2T 1.71Vdimm (OC)

Getting the stick happily running at 2400Mhz was already a blast. Yet these Veloce rams had no problem either to run at 2600Mhz HCI Memtest stable, which is kinda impressive in my book for an 8GB Dimm kit. Running them 1T was also no problem, but to keep the data aligned we decided to stick with Command Rate at 2T.

 

Here's a shot of a dummy of that infamous GeIL DBT testing machine. On the right the goal of this machine: To prevent early failure, to warrant operating speeds and test voltages and ensure proper binning of ICs.

With our i7-3770K running at 4500MHz we ripped through our RAM test suite, testing these GeIL EVO VELOCE's at 1600, 2133, 2400 and finally 2600MHz speeds. When looking at the graphs check the performance differences between the comonly used 1600MHz and this Veloce's sample box rated 2133MHz speeds. The overclocked 2400 and 2600MHz results are included as an extra. That little extra that is squeezed out of your setup and totally for free.

Let's get this party started with SuperPI 32M, a strenous test for your CPU's Integrated Memory Controller and your RAM sticks. Big scaling happening between 1600MHz and 2133MHz, close to 9.5 seconds. From there one the differences get smaller. But still the 6 secs gain at 2600MHz over the boxed speeds is pretty cool to observe. Wprime1024, being a more CPU based test, is included to show our readers that not all applications benefit from faster RAM.

 

 

The bandwith leap in the COPY and READ test over the 1600MHz kit is impressive. The Ivy Bridge platform adores high speed RAMs. However looking at the outcome in the AIDA CPU Queen test. Spot again that some applications don't benefit from the extra bandwith generated by higher RAM frequencies. The Photoworxx test on the other hand get's an impressive boost.

 

 

The Maxon Cinebench series are widely used for hardware reviews. Both the Release 10 and 11 version show very light scaling when we increase the RAM speeds.

 

 

In the encoding X264HD test, the jump from 1600 to 2133MHz is over 6 Frames Per Second. Overclocking yields some more gains, but the gains are far less once we surpass 2133Mhz.

 

 

Slight gains noticeable with added memory frequency in 3D. But nothing earthshaking to see here. To be expected as RAM speeds have hardly any influence on games either. Sometimes an increased or more stable min and average FPS are more related to the quantity, then the speed of the RAM.

 

Usually RAM vendors set the timings in the XMP profile a little looser to ensure stability of the kit. Let's see if we can tweak some to get a little extra peformance for free. Take note that we stick to the XMP voltage set and strangely enough it is 1.65Vdimm instead of the announced 1.5Vdimm on the RAM's stickers, even tough GeIL includes a little Readme leaflet. It still surprises me why they opted to choose 1.65Vdimm via XMP and warn you to adjust manually yourself. Let's take a look first at the XMP settings:

 

 

First thing is to try to see if Command Rate 1T is stable, then lower tRFC from 321 to 285. The tertiary timings and especially the big bandwith boosters tRRSR and tWWSR are set at 4 which is excellent. No more tweaking for the latter. We ended up with the below settings for the Tweaked results.

 

 

Thing we noted is that this Hynix based kits does not allow to go low on the TRFC, once we tried below 250, either the sytem refused to boot, locked up in Windows7 or 3DMark11 would crash. Let's take a result at the results:

 

Almost a 4 second lead with the tweaked RAM settings over the XMP profile in SuperPI 32M. Seems these sticks still got some room left. The AIDA bandwith results speak for themselves, not massive gains, but 600MB/s in COPY and 700MB/s in READ are pretty solid numbers, especially for the few tweaks applied.

 

 

 

 

 

Cinebench scales, though X264HD improvement is hardly noticeable. The latter benchmark loves more speed and correct applied timings. 3DMark11 puts out a better score in the PhysX subtest, yet the total score hardly budges.

 

 

 

The mandatory HCI Memtest screenshot:

Time to wrap it up: GeIL has done well with this 16GB Dual channel kit. Exceeding my expectations by far. First looks at the boxed specifications didn't seem that impressive. 2133MHz at CL10 is a hum for an enthousiast a little on the loose side. However we have to keep in mind that we are dealing with 8GB dimms and not the usual 2 or 4GB versions we frequently mess with.

 

When we rate the stock frequency, we compare it with the more comon speed of 1600MHz. Most users or vendors will skip 1333MHz for their Ivy Bridge builds. Ivy Bridge is a platform that's so versatile and it hungers for more RAM speed. Daily users will hardly see any benefit running their precious RAM sticks over 2133MHz. Yet the performance step up from 1600MHz, depending on the applications used, can be huge. GeIL labels this ram as DDR3 Hardcore Gaming Memory. Even though for gaming, speeds above 1600MHz hardly boost the rendered FPS (Frames Per Second), the overal performance of your system takes a big leap forward. Secondly with 16GB in your daily setup, close to no application will starve for memory resources. If the need for more RAM in the future is a requirement, then there are still two free DIMM slots ready to be filled up with more RAM goodness.

For those that seek a bit more then the boxed rated specs, then this is really a fun kit to play with. Boosting up the Vdimm to 1.65V we achieved a rock stable 2400MHz C10. At 1.7Vdimm 2600Mhz C11 was a no brainer either. Pretty amazing results for 8GB DIMMS. Keep in mind though that overclocking results might vary from setup to setup.

 

With the EVO VELOCE GeIL has created an impressive kit, that besides being well balanced at stock speeds, has got more than enough headroom to push that little bit more. Pretty Hardcore results in my book if i compared it to other 4GB dimms we have here at the [M]Lab. If high quantity is a must and future upgradability to even more GB's is on the top of your priority list then check out these VELOCE kits. The boxed XMP settings are well balanced and the achieved tweaking resulted in minor improvements. A perfect indication that the GeIL engineers did their homework. Hence why the Madshrimps crew happily award this Evo Veloce kit the performance award :)

 

 

PROS:

• 16GB Dual channel
• OCing headroom
• Decent XMP profile (timings)

 

CONS:

• SPD with 1.65Vdimm iso 1.5 ?
  

 

 

 

A big thanks to Jan-Hendrick from Freaks4U gaming GmbH for the Veloce review sample

Emmy and Ashley from GeIL for the support

and the Tones crew for supporting the Shrimps day in day out.