I refer you to the good site http://projectrho.com/. You will have to navigate from their to the section "Advanced Design". For assistance on why a cigar (which is actually a cylinder, so I made a mistake their a little) is better (or worse, you decide). Note on the acceleration bit- a "cigar" and a sphere will both be capable of the same final velocity (assuming they have the delta-v and comparably powerful propulsion systems). The rate at which they gain the velocity will be different.
HERE's some on why a cylinder is good/bad-
The cylinder is more aerodynamic (for take-off and landing on planets with atmospheres), and allows the use of a smaller anti-radiation shadow shield (because from the point of view of the reactor the body of the ship subtends a smaller angle). It also lends itself well to the tumbling pigeon concept since it does not have to spin as fast as a sphere of the same volume in order to generate the same centrifugal gravity.
Drawbacks include a larger surface area, and a larger "moment of inertia" for yaw and pitch maneuvers (but a lower moment of inertia for roll maneuvers). This means it takes forever to point the ship's nose in different directions as compared to a sphere, which means poor maneuverability (See short story "Hide and Seek" by Sir Arthur C. Clarke for details). Larger gyros or stronger attitude jets will be needed. A faster roll rate is actually not of much use, unless you are trying to get a weapon turret to bear on an enemy ship (See the wargame Attack Vector: Tactical for details).
Cylinder shapes are also better if your ship has a so-called "spinal mount" weapon, that is, where instead of mounting a weapon on your ship you instead build the ship around the weapon. Such weapons are typically long and skinny, which fits the profile of a cigar more than a sphere
AND now the sphere-
Spheres have the largest enclosed volume for the smallest surface area of any shape, which is a major advantage where every gram of structural mass is a penalty. They also have a smaller moment of inertia for yaw and pitch maneuvers. Drawbacks are the opposite of the cylinder: they are only slightly more aerodynamic than a brick, they don't shadow shield well, and they are lousy tumbling pigeons.
Spheres also require more internal support structure than cylinder to handle the same acceleration load, particularly if you're going to be putting decks inside of it that rely on the structural framework of the spheroidal hull for rigidity. Cylinders under acceleration support themselves in the same manner as a skyscraper building, spheres need extra bracing to keep the equator from sagging. Of course this only becomes a problem if the acceleration is greater than a tenth of a gee, neither spheres nor cylinders have any problem coping with milligee acceleration.
On the other tentacle, if the shape has to be pressurized, like a fuel tank or a crew compartment, non-spherical shapes require more bracing mass and are more expensive to construct than spherical shapes.
Ken Burnside noted that another drawback of a sphere is that your internal volume is going to have a lot of "wasted dead spaces" near the hull. Odd shaped volumes that are what happens when you have an interior wall sectioning off part of the curved surface of the sphere. Anybody who has tried to lay out a floor plan inside a Buckminster Fuller geodetic dome house knows the problem.
AND a bit of comparison-
Yet another thing to keep in mind is that using current manufacturing techniques, constructing a cylindrical hull costs about 70% of the cost of constructing a spherical hull with the same volume.
Why? Because it is more difficult to manufactured girders and plates that are bent compared to straight ones. A cylinder is constructed using straight stringers. The frames are circular, but all the frames have the same radius and radius of curvature. A sphere on the other hand uses curved stringers and circular frames all of different sizes (well, there are actually two frames of each given radius, but you understand the point I'm trying to make).
On most modern wet-navy warships, the hull plates are mostly straight, with a few bent in one dimension, and only a couple bent spherically in two dimensions. Bending is expensive. Eliminating the bending cost will require one and perhaps two breakthroughs in manufacturing technology.
SO there you have it. (capitalized 1st words are for differentiation of sections). EDIT: Forgot to mention this- the shapes are reduced to the fundamentals. As for the exhaust velocity- I read it somewhere and cannot seem to find the source. I am reasonably confident that it is on the Atomic Rockets site.